DEV Community: UplinkSwitch

Best AI chat with no restrictions, checked (November 2026)

UplinkSwitch — Tue, 03 Nov 2026 00:46:15 +0000

"No restrictions" is a claim, not a feature, so this ranking uses things you can check yourself in a minute: can you type before you sign up, what sign-up asks for, the smallest amount that keeps you going, which models are named, and where the conversation ends up. Each site's line was read from its own pricing or home page on November 2, 2026 and is linked below. The Unrestrictable line comes from the site itself and from its daily model check, which asks every build a question the stock model refuses.

The top four, side by side

What you get	Unrestrictable	Venice	FreedomGPT	aiwithnorestrictions
Free chat with no account	3 turns a day	Not stated; sign-up offered	Yes, no cap stated	Yes, no daily cap
Free tier once signed in	10 messages, 5 images, 3 files a day	10 text prompts, 15 image prompts a day	100 bonus credits	Same as without
Sign-up asks for	Email, Google or GitHub	Sign-up offered for the free tier	Account for Pro only	Account for a plan only
Lowest price to keep going	$10 top-up, no plan	$18 a month	$20 a month	Plus and Pro, no price shown
Chat models	145 uncensored builds	Named open-weight models	400+ models, hosted	Own open-weight models, unnamed
Where chats are stored	In your browser	In your browser	History on Pro	Stateless, not stated
Published limits	Acceptable use page	Terms	Model and service rules apply	Restrictions page, 99.9% claim

1. Unrestrictable

Three free turns a day with no account, on 87 of the 145 chat models. Sign-in is an email address, Google or GitHub: no phone, no ID. After the free turns it is a top-up from $10 with no plan, spent per token on any model, or plans at $18, $68, $200 a month. Chats stay in your browser; prompts pass through and are not written anywhere, with one usage record per request for billing. System prompt, temperature, file reading, web search and characters are on the free tier. The limits are on one page: anything that sexualizes a minor is refused before it reaches a model, and you must be 18 or older.

2. Venice

The best-known name on this list. Its pricing page gives the free tier 10 text prompts and 15 image prompts a day, and Pro at $18 a month adds unlimited text, 1,000 images a day, custom characters, longer context and encrypted chat backup. Conversations are stored locally. The free tier is offered through sign-up, and there is no way to pay less than the monthly plan for the chat.

3. FreedomGPT

Free to try with no card and no account, per its home page, with automatic model selection, web search and image generation on the free plan and 100 bonus credits when you make an account. Pro is $20 a month for 2,200 credits, premium models, video and conversation history. It names 400+ hosted models, so uncensored builds sit beside the mainstream ones, and its own page says model and service rules still apply to the uncensored options.

4. aiwithnorestrictions.com

The page that ranks first on Google for the phrase. It states no daily message cap, no account, no email and no country block for the free chat, on open-weight models it hosts itself. The trade: the free chat sends the model about 5,000 tokens of the thread, so long conversations lose their start, and the free tier shows promotions. Plus and Pro are mentioned without prices on the home page, the operator is not named, and a footnote qualifies the headline to 99.9% with a restrictions page.

5. notrack.ai

Free, no account, adults only, and candid about it: the home page says the limits that exist are published, not hidden, and claims about 99.9% of questions are answered directly. It runs one model of its own, NoTrack, with other models in the full chat. Sessions are stateless and nothing trains a model. Plus and Pro add longer memory and no waiting at busy times, prices not on the home page.

6. Shannon AI

An uncensored assistant with its own model, Shannon 3.1, beside 12 open-weight models, memory and web search. Its plans start at $5.99 a month and go to $19.99 and $49.99, with crypto buyers getting two months for one. The free tier is 30,000 tokens and three web searches a day, and it asks you to sign in before the first message. Ranked here for the sign-in wall and the plan-only pricing, not for the model.

7. A model on your own machine

Ollama or LM Studio with an abliterated build from Hugging Face is free forever and never leaves your computer. It needs a GPU with 8 GB of memory for the small builds and far more for the ones that feel like a hosted chat, and the small ones are noticeably less capable. The hosted sites above exist because most people do not want to run this.

What the ranking does not measure

How good the answers are. That depends on the model you pick more than the site, and the same abliterated builds appear on several of these. Pick the site by the terms above, then pick the model: every one on Unrestrictable is listed with its price per million tokens.

Questions

Which AI chat has no restrictions? None of them has none. Every hosted chat has a line somewhere, and the honest ones publish it. The seven here are the ones whose models have the stock refusal layer removed or tuned out, ranked on free use, sign-up, price and where chats are stored.

Is there a free AI chat with no restrictions and no sign-up? Yes. On Unrestrictable, three free turns a day on 87 models with no account. aiwithnorestrictions.com and notrack.ai also open without an account. Venice and FreedomGPT give a free tier once you sign up.

What is the cheapest way to keep going after the free turns? A top-up from $10 with no plan on Unrestrictable, spent per token across 145 chat models. Elsewhere it is a plan: Shannon AI from $5.99 a month, Venice $18, FreedomGPT $20.

Do these chats store my conversations? Unrestrictable and Venice keep chats in your browser. aiwithnorestrictions.com and notrack.ai say sessions are stateless and no profile is built. FreedomGPT Pro adds conversation history. Shannon AI keeps memory as a feature.

Who wrote this list? Unrestrictable did, so read the first entry as the house pick. Every other line links to the page it was read from, dated November 2, 2026.

Originally published at unrestrictable.ai/best-ai-chat-no-restrictions; the model count and prices there are read from the live catalog.

RisuAI API settings for a custom OpenAI-compatible API

UplinkSwitch — Mon, 02 Nov 2026 06:33:53 +0000

Checked 2026-11-02

Point RisuAI's Custom API at RelaySwitch: four fields, the model id to type, and what each message costs.

The RisuAI API settings for a hosted model live in one panel: pick Custom API as the model, paste https://api.relayswitch.io/v1 in URL, your key in Key/Password, and a model id in Request Model. RisuAI calls it a custom API; older builds and most forum posts call the same thing a reverse proxy. One key reaches every model in the catalog, from the uncensored roleplay builds to Claude, billed per token from a prepaid balance.

What you need

RisuAI in the browser, on desktop, or the Android or iOS app.
A RelaySwitch key from the console (email code sign-in). Keys start with sk-relay-.
A balance. Top-ups start at $10, by card or crypto, with no subscription. Before the first top-up the console chat runs 12 free turns a day, so you can try a model there before you connect it.

Where the RisuAI API settings are

Open the menu (the hamburger icon, top left) and choose Settings.
Open Chat Bot.
Click Model to open the model picker.
Choose Custom API. Older releases list it as Reverse Proxy or OpenAI Reverse proxy.

The fields

The Custom API block shows these fields, in this order.

text

1URL: https://api.relayswitch.io/v1

2Key/Password: sk-relay-...

3Request Model: glm-5.3-uncensored

4Format: OpenAI Compatible

5Response Streaming: on

URL: https://api.relayswitch.io/v1, with the /v1.
Key/Password: your sk-relay- key. The placeholder reads "leave it blank if it hasn't password"; ours always needs the key.
Request Model: the exact id from the Models page, not the display name. RisuAI does not fetch a model list, so type it.
Format: OpenAI Compatible (the default) for every model. The other options are OpenAI Response API, Anthropic Claude, Mistral, Google Cloud and Cohere.
Response Streaming: on, so the reply types out as it arrives. Every model streams.

The short URL works because Autofill Request URL (Settings > Advanced Settings) is on by default: RisuAI adds /chat/completions to a URL that ends in v1. If you turned it off, paste the full endpoint, https://api.relayswitch.io/v1/chat/completions.

Anthropic format

For a Claude model you can switch Format to Anthropic Claude. Keep the same URL; with Autofill Request URL on, RisuAI adds /messages (full endpoint with it off: https://api.relayswitch.io/v1/messages). The key stays in Key/Password. RisuAI sends it as an x-api-key header with anthropic-version: 2023-06-01, and the API accepts that header as well as Authorization: Bearer.

text

1URL: https://api.relayswitch.io/v1

2Key/Password: sk-relay-...

3Request Model: claude-sonnet-4-6

4Format: Anthropic Claude

13 Claude ids answer the Anthropic format today, including claude-sonnet-4-6, claude-opus-5-5, claude-fable-5. Every other model on this page stays on OpenAI Compatible. The OpenAI format reaches the Claude ids too, so Anthropic Claude is optional: pick it when you want the request in Claude's own format.

Which model to type

Uncensored and roleplay builds first: they answer without the stock refusals, and the roleplay ones are tuned on fiction and dialogue. Prices are per million tokens. The turn columns price one 2K-in / 300-out turn: 2,000 tokens of card, persona and recent history in, 300 tokens of reply out.

Model	Request Model	Context	Input / M	Output / M	one 2K-in / 300-out turn	100 turns
GLM-5.3 Flash Uncensored	`glm-5.3-flash-uncensored`	not stated	$0.1086	$0.3802	$0.00033	$0.0331
GLM-5.3 Uncensored	`glm-5.3-uncensored`	not stated	$0.291	$0.873	$0.00084	$0.0844
Gemma 4 31B Fabled	`gemma-4-31b-fabled`	262K tokens	$0.35	$1.575	$0.0012	$0.1173
Gemma 4 31B Novelist	`gemma-4-31b-novelist`	262K tokens	$0.35	$1.575	$0.0012	$0.1173
Qwen3.8 27B Novelist	`qwen3.8-27b-novelist`	262K tokens	$0.35	$2.10	$0.0013	$0.133
Cydonia 24B v4.3	`cydonia-24b-v4.3`	131K tokens	$0.484	$0.60	$0.0011	$0.1148
Uncensored AI Roleplay	`uncensored-ai-roleplay`	128K tokens	$0.60	$2.34	$0.0019	$0.1902
DeepSeek V4.1 Flash Abliterated	`deepseek-v4.1-flash-abliterated`	not stated	$0.875	$1.75	$0.0023	$0.2275
DeepSeek V4 Pro Uncensored	`deepseek-v4-pro-uncensored`	not stated	$1.00	$2.00	$0.0026	$0.26
MythoMax L2 13B	`mythomax-l2-13b`	4K tokens	$0.404	$0.404	$0.00093	$0.0929

USD, read from the RelaySwitch catalog when this page was built.

The flagships, for longer memory and stronger writing:

Model	Request Model	Context	Input / M	Output / M	one 2K-in / 300-out turn	100 turns
Claude Sonnet 4.6	`claude-sonnet-4-6`	1M tokens	$1.80	$9.00	$0.0063	$0.63
Claude Opus 5.5	`claude-opus-5-5`	1M tokens	$2.80	$10.00	$0.0086	$0.86
Claude Fable 5	`claude-fable-5`	1M tokens	$3.00	$15.00	$0.0105	$1.05
Gemini 3.1 Pro Preview	`gemini-3.1-pro-preview`	1M tokens	$1.20	$7.20	$0.0046	$0.456
DeepSeek V3.2	`deepseek-v3.2`	not stated	$1.08	$1.6801	$0.0027	$0.2664
Grok 4.6	`grok-4.6`	500K tokens	$0.60	$1.80	$0.0017	$0.174
GLM 5.3	`glm-5.3`	1M tokens	$0.3276	$1.0296	$0.00096	$0.0964
GPT 5	`gpt-5`	400K tokens	$2.50	$20.00	$0.011	$1.10

USD, read from the RelaySwitch catalog when this page was built.

The uncensored models page lists every build the catalog tags, and the roleplay chat runs character cards in the browser on the same key.

What a long chat costs

Every message resends the card and as much history as RisuAI's Max Context Size allows, so input grows as the chat does until it hits that cap. Per message at three context sizes, with a 300-token reply:

Context sent	Gemma 4 31B Fabled	GLM-5.3 Uncensored	Claude Sonnet 4.6
4,000 tokens	$0.0019	$0.0014	$0.0099
8,000 tokens	$0.0033	$0.0026	$0.0171
16,000 tokens	$0.0061	$0.0049	$0.0315

USD per message, read from the RelaySwitch catalog when this page was built.

100 messages at 16,000 tokens of context: $0.6073 on Gemma 4 31B Fabled, $0.4918 on GLM-5.3 Uncensored, $3.15 on Claude Sonnet 4.6.

Keep Max Context Size at or under the model's window in the Context column; MythoMax L2 13B stops at 4K tokens. The console's Usage page lists every request with its tokens and charge, and each one has a signed receipt.

On Android and iOS

The app has the same settings panel as the web build: menu > Settings > Chat Bot > Model > Custom API, then the same URL, Key/Password, Request Model and Format. Paste the key from the console rather than typing it; a stray space or a missing character answers with a 401. Streaming works the same way on mobile.

Errors

RisuAI has no Connect or Test button, so the first message is the test. An error shows in the chat in place of the reply.

401: the key is wrong or revoked. Paste it again in Key/Password.
402: the balance is empty, or the key reached its own budget. Top up or raise the budget in the console.
403 scope_denied: the key is limited to a list of models and this id is not on it. Add it to the key or use another key.
404 model_not_found: Request Model is not an id the catalog lists. Copy it from the Models page. route_not_found: the URL is wrong; with Autofill Request URL off, it must end in /chat/completions (or /messages on the Anthropic format).
429: the key hit its requests or tokens per minute limit. Wait a few seconds and resend.
503: no host could serve that model right now. Resend, or pick another model.

The API docs list every error code.

Sources

Field names, the Format options, the model picker path and the Autofill Request URL behavior are from RisuAI's repository on the main branch, read 2026-11-02: src/lang/en.ts, BotSettings.svelte, openAI/requests.ts, anthropic.ts and database.svelte.ts (Autofill Request URL defaults to on). The documentation site, docs.risuai.net, was unreachable on 2026-11-02, and the GitHub wiki is marked outdated by its maintainers.

Questions

What are the best RisuAI API settings?

Custom API as the model, URL https://api.relayswitch.io/v1, your key in Key/Password, Format OpenAI Compatible, Response Streaming on, and an id from the Models page in Request Model. Then set Max Context Size at or under that model's window; it decides most of what a message costs.

Where are the RisuAI API settings on Android?

In the same place as on the web: menu > Settings > Chat Bot > Model > Custom API. The Android and iOS app share the settings panel with the browser build, so the URL, Key/Password, Request Model and Format fields are identical.

Are the RisuAI API settings in Reddit posts still right?

The fields are, the names are not. Older posts call the option Reverse Proxy or OpenAI Reverse proxy; current builds call it Custom API in the model picker. URL, Key/Password and Request Model work the same way.

Is the RisuAI wiki on GitHub up to date?

No. The GitHub wiki is marked outdated, and docs.risuai.net was unreachable on 2026-11-02. The field names on this page come from the app source on the main branch, read the same day.

What URL goes in RisuAI custom API?

https://api.relayswitch.io/v1. With Autofill Request URL on (the default), RisuAI adds /chat/completions itself, or /messages on the Anthropic Claude format. With it off, paste https://api.relayswitch.io/v1/chat/completions.

Does RisuAI work with an OpenAI-compatible proxy?

Yes. Custom API on the OpenAI Compatible format is RisuAI's proxy setting. Any endpoint that speaks the Chat Completions format works there, including RelaySwitch.

How do I test the connection in RisuAI?

Send a message. RisuAI has no Test button and does not fetch a model list, so the first reply is the test. An error code in the chat tells you which field to fix: 401 the key, 404 the model id or the URL.

Can I use Claude in RisuAI with a custom API?

Yes. Type a Claude id in Request Model and leave Format on OpenAI Compatible, or switch Format to Anthropic Claude with the same URL and key. RelaySwitch accepts the x-api-key header RisuAI sends on that format.

How much does a RisuAI message cost?

It depends on the model and how much history goes with the message. One 2K-in / 300-out turn on GLM-5.3 Flash Uncensored costs $0.00033 today, or $0.0331 for 100. A larger Max Context Size raises the input side once the chat fills it.

Every model in one RisuAI connection

Prepaid from $10 by card or crypto. No KYC, no subscription.

Models Get a key

Originally published at relayswitch.io/guides/risu; the prices there are read from the live catalog.

Chub AI API key and reverse proxy setup

UplinkSwitch — Mon, 02 Nov 2026 06:28:21 +0000

Checked 2026-11-02

Secrets → OpenAI → Reverse Proxy with the RelaySwitch base URL and key, the model field under Configuration, what one 2K-in / 300-out turn costs on each model, and the fix for an empty response.

A Chub AI API key goes in Secrets, the panel behind the three-line menu of any chat, and the model goes in Configuration. For a key from another provider Chub uses its Reverse Proxy entry under OpenAI: a base URL and a key. The base URL for RelaySwitch is https://api.relayswitch.io/v1, ending in /v1, because Chub adds /chat/completions itself.

One key runs every chat model on the Models page, the uncensored and roleplay builds included, from a prepaid balance that starts at $10 by card or crypto. No subscription, no KYC. Prices on this page are read from the live catalog at render.

Where the Chub AI API key goes: Secrets → OpenAI → Reverse Proxy

Create a key in the console. It is shown once; keys start with sk-relay-.
On chub.ai, open any chat and click the three-line menu at the top right. Choose Secrets.
Expand OpenAI. Set OpenAI Access API to Reverse Proxy (not My Own API Key, which is for a key issued by OpenAI).
OpenAI Reverse Proxy URL: https://api.relayswitch.io/v1. Keep the /v1, add nothing after it.
Reverse Proxy Key: your key.
Click Check Proxy. Chub reads /v1/models from the URL with the key and answers Valid API Key/Proxy, API key is invalid (a 401: paste the key again) or Network Error. Try again later! (the URL is wrong, usually a missing /v1 or a full /chat/completions path).

text

1Menu (three lines, top right) → Secrets → OpenAI

2OpenAI Access API: Reverse Proxy

3OpenAI Reverse Proxy URL: https://api.relayswitch.io/v1

4Reverse Proxy Key: sk-relay-...

5→ Check Proxy

7Menu → Configuration → Prompt Structure → API/Model

8API: OpenAI / Proxy

9Model: glm-5.3-flash-uncensored

Configuration → Prompt Structure → API/Model

Open the menu again and choose Configuration. Pick or create a preset.
Under Prompt Structure, open API/Model and set the API to OpenAI / Proxy.
Set the model. The dropdown is filled from the list Check Proxy fetched, and it fills inconsistently: if it is empty or stale, refresh the page, open the preset again, and type the id exactly, such as glm-5.3-flash-uncensored.
In the same preset, set max tokens to 1,000 or more for a model that thinks before it answers, and turn on text streaming if you want replies to appear as they are written.
Save, go back to the chat and send a message. The first reply is the real test; it bills like any other.

Assistant Prefill and Post History Instructions in the preset are sent with every turn and bill as input tokens. Keep them short on a long chat.

Which model to type

The 10 builds the catalog tags for uncensored or roleplay use come first, then the flagships on the same key. Prices are USD per million tokens; the last two columns price one 2K-in / 300-out turn and 100 of them. A $10 top-up covers about 30,179 such turns on GLM-5.3 Flash Uncensored.

Model	Model field	Input	Output	one 2K-in / 300-out turn	100 turns
GLM-5.3 Flash Uncensored	`glm-5.3-flash-uncensored`	$0.1086	$0.3802	$0.00033	$0.0331
GLM-5.3 Uncensored	`glm-5.3-uncensored`	$0.291	$0.873	$0.00084	$0.0844
Gemma 4 31B Fabled	`gemma-4-31b-fabled`	$0.35	$1.575	$0.0012	$0.1173
Gemma 4 31B Novelist	`gemma-4-31b-novelist`	$0.35	$1.575	$0.0012	$0.1173
Qwen3.8 27B Novelist	`qwen3.8-27b-novelist`	$0.35	$2.10	$0.0013	$0.133
Cydonia 24B v4.3	`cydonia-24b-v4.3`	$0.484	$0.60	$0.0011	$0.1148
Uncensored AI Roleplay	`uncensored-ai-roleplay`	$0.60	$2.34	$0.0019	$0.1902
DeepSeek V4.1 Flash Abliterated	`deepseek-v4.1-flash-abliterated`	$0.875	$1.75	$0.0023	$0.2275
DeepSeek V4 Pro Uncensored	`deepseek-v4-pro-uncensored`	$1.00	$2.00	$0.0026	$0.26
MythoMax L2 13B	`mythomax-l2-13b`	$0.404	$0.404	$0.00093	$0.0929
Claude Sonnet 4.6	`claude-sonnet-4-6`	$1.80	$9.00	$0.0063	$0.63
Claude Opus 5.5	`claude-opus-5-5`	$2.80	$10.00	$0.0086	$0.86
Claude Fable 5	`claude-fable-5`	$3.00	$15.00	$0.0105	$1.05
Gemini 3.1 Pro Preview	`gemini-3.1-pro-preview`	$1.20	$7.20	$0.0046	$0.456
DeepSeek V3.2	`deepseek-v3.2`	$1.08	$1.6801	$0.0027	$0.2664
Grok 4.6	`grok-4.6`	$0.60	$1.80	$0.0017	$0.174
GLM 5.3	`glm-5.3`	$0.3276	$1.0296	$0.00096	$0.0964
GPT 5	`gpt-5`	$2.50	$20.00	$0.011	$1.10

USD, read from the RelaySwitch catalog when this page was built. A turn here is 2,000 tokens in (system prompt, character and recent history) and 300 out.

Chub resends the definition and as much history as fits with every turn, so the input side grows with the chat. The same 300-token reply at three history sizes:

Tokens sent per turn	GLM-5.3 Flash Uncensored	Uncensored AI Roleplay	Claude Sonnet 4.6
4,000	$0.00055	$0.0031	$0.0099
8,000	$0.00098	$0.0055	$0.0171
16,000	$0.0019	$0.0103	$0.0315

USD per turn at live catalog rates. Cached input tokens, when a model reports them, bill at its cached rate.

The uncensored models page lists every tagged build with its daily check, and the roleplay chat runs a character card on the same key in the browser, so you can try a model before typing it into Chub.

Empty response received from API, and the other errors

Work through these in order; the first is the usual cause.

Empty response received from API. The model returned no reply text. Most often a model that thinks before it answers spent a small max tokens budget on thinking: raise max tokens in the preset to 1,000 or more, or switch to a build without a thinking step. The other cause is a 402: the balance or the key's budget ran out, and Chub shows the empty reply rather than the code. Check the balance in the console.
401: the key is wrong or revoked. Paste it again into Reverse Proxy Key and click Check Proxy.
402: the prepaid balance is too low, or the key reached its budget. Top up or raise the budget; the next turn goes through.
403: the key is narrowed to a model list and the model field names another one. Add the model to the key or change the field.
404: either the model id is misspelled or gone from the catalog (model_not_found), or the URL ends in /chat/completions so Chub posted to a path that does not exist (route_not_found). Use an id from the Models page and https://api.relayswitch.io/v1 exactly.
429: the key hit its per-minute request or token limit. Wait a few seconds, or raise the limit on the key.
Network Error. Try again later! after Check Proxy: Chub could not reach /v1/models at the URL. Fix the URL, not the key.
Old messages forgotten: the preset's context size is below the model's window. Raise it; the long-chat table shows what that costs per turn.

To keep a key problem and a Chub setting apart, test outside Chub first. List the models the key can call:

bash

1curl https://api.relayswitch.io/v1/models -H "Authorization: Bearer sk-relay-..."

Then send one short turn on the id you picked:

bash

1curl https://api.relayswitch.io/v1/chat/completions \

2 -H "Authorization: Bearer sk-relay-..." \

3 -H "Content-Type: application/json" \

4 -d '{"model": "glm-5.3-flash-uncensored", "max_tokens": 1000,

5 "messages": [{"role": "user", "content": "Say hello in one sentence."}]}'

The reply is in choices[0].message.content. If it comes back, the key, the URL and the id are right and whatever is left is a Chub setting. The API docs list every error code.

Sources

Chub's own API Connections page names the providers Secrets accepts and the OpenAI Access API toggle but does not document the Reverse Proxy fields. The field labels, the Check Proxy results and the model-list quirk are from two community setup guides, Chub Secrets and Config and the Chub AI setup page, all read 2026-11-02. Chub also sells its own models by subscription; that route needs no key and is not covered here. Prices and model ids come from the RelaySwitch catalog at render. The same key runs JanitorAI (full /chat/completions URL), SillyTavern and RisuAI (base URL, like Chub).

Questions

Is there a free API for Chub AI?

Not from this key: it is prepaid from $10 with no subscription, and that first top-up covers about 30,179 turns of 2,000 tokens in and 300 out on GLM-5.3 Flash Uncensored. Before the first top-up the console chat runs 12 free turns a day, enough to try a model before pasting it into Chub.

How do I get a free AI API key?

Providers that hand out free keys cap them hard or end the free tier, which is how most people arrive at a reverse proxy. Here a key is created in the console in a minute and runs on a prepaid balance; 12 free console turns a day let you test a model first.

Which API is best for Chub AI?

For a provider key, the OpenAI Reverse Proxy entry in Secrets, because it takes any OpenAI-compatible base URL and key. Which model to put behind it depends on budget and taste: one 2K-in / 300-out turn on GLM-5.3 Flash Uncensored costs $0.00033 today, and the table above prices the flagships beside it.

How can I get my API key?

Sign in to the console with an email code, add a balance, and create a key. It is shown once; paste it into Reverse Proxy Key in Chub's Secrets. You can narrow the key to a model list, give it a USD budget and per-minute limits, and revoke it any time.

Where do I put my API key in Chub AI?

Open a chat, click the three-line menu at the top right, choose Secrets, expand OpenAI, set OpenAI Access API to Reverse Proxy, and paste the key into Reverse Proxy Key with the base URL in OpenAI Reverse Proxy URL. Click Check Proxy.

What proxy URL do I use for Chub AI?

https://api.relayswitch.io/v1. It ends in /v1 and nothing more; Chub adds /chat/completions itself, so a full URL gets a 404.

Why does Chub say empty response received from API?

The model sent back no text. Raise max tokens in the preset to 1,000 or more, since a thinking model can spend a small budget before it writes the reply, or pick a build without a thinking step. If that does not fix it, check the balance: an empty balance or key budget shows up the same way. The curl check above prints the code that names the cause.

What are the Chub AI API settings for a reverse proxy?

Secrets holds the connection (OpenAI Access API: Reverse Proxy, OpenAI Reverse Proxy URL https://api.relayswitch.io/v1, Reverse Proxy Key). Configuration holds the preset: API/Model under Prompt Structure, then max tokens, context size, streaming, Assistant Prefill and Post History Instructions.

Does the same key work in JanitorAI and SillyTavern?

Yes. SillyTavern and RisuAI take the base URL https://api.relayswitch.io/v1 like Chub. JanitorAI takes the full https://api.relayswitch.io/v1/chat/completions as its Proxy URL. The key, the balance and the model ids are the same everywhere.

One reverse proxy entry, every model

Prepaid from $10 by card or crypto. No KYC, no subscription.

Uncensored models Get a key

Originally published at relayswitch.io/guides/chub; the prices there are read from the live catalog.

SillyTavern LLM settings for an OpenAI-compatible API

UplinkSwitch — Mon, 02 Nov 2026 06:22:49 +0000

Checked 2026-11-02

Connect SillyTavern to RelaySwitch as a Custom (OpenAI-compatible) Chat Completion source, then set the four samplers that reach the model, a Context Size that fits it, and reasoning.

Most SillyTavern LLM settings guides assume a local backend, where every sampler on the panel goes to the model. A hosted OpenAI-compatible API works differently. SillyTavern sends a short list of standard parameters, hides the rest, and leaves Context Size for you to set by hand.

This guide connects SillyTavern to https://api.relayswitch.io/v1, shows exactly which settings travel with each message, and sets Context Size per model from the live catalog. Every message bills per token from a prepaid balance, so the last sections price a message on each model in the table.

Looking for a SillyTavern proxy? For a model backend, that is the Custom (OpenAI-compatible) source below: a URL and a key that answer in the OpenAI format, or the Claude source's Proxy Server URL for the Anthropic format. It is not requestProxy in config.yaml, which sends SillyTavern's own outgoing traffic through a network proxy, and not the Reverse proxying page in the admin docs, which puts Traefik in front of SillyTavern for remote access.

What you need

SillyTavern installed and running. The steps follow the release branch.
A RelaySwitch key from the console (email code sign-in). Keys start with sk-relay-.
A balance. Top-ups start at $10, by card or crypto, with no subscription. Before the first top-up the console chat runs 12 free turns a day, so you can try a model before connecting it here.

Connect the Custom source

Open API Connections (the plug icon in the top bar).
Set API to Chat Completion.
Set Chat Completion Source to Custom (OpenAI-compatible).
Custom Endpoint (Base URL): https://api.relayswitch.io/v1. Keep the /v1. SillyTavern adds the rest; its docs put it plainly: "Do NOT add the /chat/completions suffix."
Custom API Key: your key. SillyTavern hides it after you connect.
Click Connect. The Available Models dropdown fills from /v1/models. Pick a model there, or type its id in Enter a Model ID.
Click Test Message. It sends one short request, which bills like any other.

text

1API: Chat Completion

2Chat Completion Source: Custom (OpenAI-compatible)

3Custom Endpoint (Base URL): https://api.relayswitch.io/v1

4Custom API Key: sk-relay-...

5Enter a Model ID: claude-sonnet-5-5

Use the id the Models page prints, such as claude-sonnet-5-5, not the display name.

The Anthropic format: Claude source with a reverse proxy

SillyTavern can also send the Anthropic Messages format, through its Claude source. The reverse proxy block sits in the same connection panel; SillyTavern's docs do not cover it, so the labels here are from the UI.

Set Chat Completion Source to Claude.
Proxy Server URL: https://api.relayswitch.io/v1. SillyTavern adds /messages.
Proxy Password: your sk-relay- key. The field's note says it is used as the password for the proxy instead of an API key, and the Claude API key field above it is ignored once a proxy is set.
Pick the model in Claude Model, then click Connect.

text

1API: Chat Completion

2Chat Completion Source: Claude

3Proxy Server URL: https://api.relayswitch.io/v1

4Proxy Password: sk-relay-...

5Claude Model: claude-sonnet-4-6

Claude Model is a fixed dropdown. These 9 of its ids answer the Anthropic format in the catalog today:

Model	Id	Input	Output	Per message
Claude Fable 5	`claude-fable-5`	$3.00	$15.00	$0.0135
Claude Haiku 4.5	`claude-haiku-4-5`	$0.60	$3.00	$0.0027
Claude Haiku 4.5	`claude-haiku-4-5-20251001`	$0.60	$3.00	$0.0027
Claude Opus 4.6	`claude-opus-4-6`	$1.80	$7.50	$0.0077
Claude Opus 4.7	`claude-opus-4-7`	$1.80	$7.50	$0.0077
Claude Opus 4.8	`claude-opus-4-8`	$1.80	$7.50	$0.0077
Claude Opus 5	`claude-opus-5`	$1.80	$7.50	$0.0077
Claude Sonnet 4.6	`claude-sonnet-4-6`	$1.80	$9.00	$0.0081
Claude Sonnet 5	`claude-sonnet-5`	$1.20	$3.00	$0.0045

USD per million tokens, read from the RelaySwitch catalog when this page was built. Per message: 3,000 tokens in, 300 out.

The dropdown has no entry for claude-fable-5.1, claude-haiku-5-5, claude-opus-5-5, claude-sonnet-5-5. Reach those through the Custom source, where any id can be typed.

Prompt Post-Processing

Prompt Post-Processing, in the connection panel under the Custom source, reshapes the message list before it leaves SillyTavern. The options, as the panel lists them:

None
Merge consecutive roles (with tools)
Semi-strict (alternating roles; with tools)
Strict (user first, alternating roles; with tools)
Merge consecutive roles (no tools)
Semi-strict (alternating roles; no tools)
Strict (user first, alternating roles; no tools)
Single user message (no tools)

Leave it on None unless a model complains. The merge options join back-to-back messages from the same role; Semi-strict also keeps user and assistant turns alternating. Pick Strict when a model answers a 400 about roles or message order: a character chat opens on the card's greeting, an assistant message, and Strict puts a user message first. The "no tools" variants drop function calling; keep a "with tools" one if you use tool calls.

SillyTavern LLM settings that reach the model

With a Custom source, the AI Response Configuration panel shows Temperature, Frequency Penalty, Presence Penalty, Top P, Seed and Max Response Length (tokens). Those are the standard Chat Completions parameters. SillyTavern's server builds this body for every message:

json

1{

2 "model": "claude-sonnet-5-5",

3 "messages": [ ... ],

4 "temperature": 1,

5 "top_p": 1,

6 "frequency_penalty": 0,

7 "presence_penalty": 0,

8 "max_tokens": 300,

9 "stream": true

10}

Custom stopping strings add a stop list, a Seed of 0 or more adds seed, and Multiple swipes per generation adds n. The API forwards all of these to the model unchanged. A model that refuses one answers with a 400 error.

Top K, Min P, Top A, Repetition Penalty, DRY, XTC, Mirostat and the other Text Completion samplers do not appear for a Custom source, and SillyTavern does not send them. Settings you made for a local backend have no effect here. To send one anyway, open Additional Parameters in the connection panel and add it under Include Body Parameters as YAML:

yaml

1min_p: 0.05

2repetition_penalty: 1.05

3reasoning_effort: high

The API passes extra fields through as well. Whether a model applies min_p or repetition_penalty depends on the model; the four standard samplers are the ones to tune first. To stop SillyTavern sending a field a model rejects, list it under Exclude Body Parameters:

yaml

1- frequency_penalty

2- presence_penalty

Starting points for long-form fiction: Temperature 0.8 to 1.0, Top P 0.95, Frequency Penalty 0.1 to 0.3 against repeated phrases, Presence Penalty 0 to 0.3 to push new topics. Change one at a time and swipe a few replies before judging.

Context Size per model

Context Size (tokens) is the most SillyTavern sends as the prompt: card, system prompt and chat history, minus the response length. A fresh install starts Chat Completion at 4,095 tokens of context and 300 response tokens. On a hosted model with a far larger window, that drops older messages long before the model needs it to.

SillyTavern does not read a context window from a Custom source. The slider goes past any model's limit, so set it from the Context column in the table below. Set it too high and a long chat fails once the prompt passes the model's window; Euryale 70B v2.3 stops at 20K tokens.

Context Size is also the cost dial. Once a chat fills it, every message sends that many input tokens, so a 32K setting costs several times more per message than a short chat on the same model. The costed example below prices both.

Reasoning

Request model reasoning in the AI Response Configuration panel shows a model's reasoning as a collapsible block above the reply when the model returns it in a separate field. For a model that writes its reasoning inline between <think> tags, turn on Auto-Parse under Advanced Formatting > Reasoning instead.

Reasoning Effort on a Custom source is sent only for model ids SillyTavern knows from OpenAI, such as gpt-5.5. For any other model, put reasoning_effort: high (or low, medium) under Include Body Parameters. The API passes it to the model, and models with an effort setting use it.

Reasoning tokens are output tokens. They count against Max Response Length and bill at the output rate, even when the block stays collapsed. If replies come back empty or cut off, raise Max Response Length; SillyTavern's docs suggest 1024 to 4096 for reasoning models.

Which models to pick

Four general models and 2 the catalog tags for fiction and character dialogue (Gemma 4 31B Fabled, Euryale 70B v2.3). Prices are per million tokens. The message columns assume 3,000 tokens in (card plus recent history) and 300 out, the default response length.

Model	Context	Input	Output	Per message	100 messages
Claude Sonnet 5.5	1M tokens	$1.20	$3.00	$0.0045	$0.45
GPT-5.5	1.1M tokens	$1.80	$9.00	$0.0081	$0.81
DeepSeek V4 Pro 0423	1M tokens	$0.792	$2.376	$0.0031	$0.3089
GLM 5.3	1M tokens	$0.3276	$1.0296	$0.0013	$0.1292
Gemma 4 31B Fabled	262K tokens	$0.35	$1.575	$0.0015	$0.1523
Euryale 70B v2.3	20K tokens	$1.972	$1.972	$0.0065	$0.6508

USD, read from the RelaySwitch catalog when this page was built. Context is the developer's stated window; set Context Size at or under it.

The general models carry the largest context windows. The fiction-tuned builds are trained on stories and dialogue and have smaller windows, so check the Context column before you raise Context Size. The roleplay chat runs character cards in the browser on the same key, and the character card guide explains what each card field costs in tokens.

What a chat costs

Early in a chat on Claude Sonnet 5.5, a message sends about 3,000 tokens and gets 300 back: $0.0045 each, or $0.45 for 100 messages.

Later, with Context Size at 32K and the history filling it, each message sends 32,000 tokens: $0.0393 each, or $3.93 for 100. The same 100 full-context messages on GLM 5.3 come to $1.08.

Swipes, Continue and Test Message are requests too, each billed the same way. When the model reports cached input tokens, they bill at its cached rate, which lowers the cost of a long chat that resends the same history. The console's Usage page lists every request with its tokens and charge. The pricing page has the full billing rules.

Errors and fixes

List the models your key can call, from the machine that runs SillyTavern:

bash

1curl https://api.relayswitch.io/v1/models -H "Authorization: Bearer sk-relay-..."

401: the key is wrong or revoked. Paste it again in Custom API Key and click Connect.
402: the balance is too low. Top up in the console; the next message goes through.
404: the model id is not in the catalog. Pick it from Available Models instead of typing it.
429: the key hit its rate limit. Wait a few seconds before the next swipe.
A 400 naming a parameter: the model refuses that field. Add it to Exclude Body Parameters, or remove it from Include Body Parameters.
A 400 about roles or message order: some models want alternating user and assistant turns. Set Prompt Post-Processing in the connection panel to Semi-strict, or Strict if the first message must be the user's.
Connection fails or the model list stays empty: the URL is missing /v1 or ends in /chat/completions. Use https://api.relayswitch.io/v1 exactly.
Empty or cut-off replies: a reasoning model used up Max Response Length before it wrote the reply. Raise it.
Old messages forgotten: Context Size is still at 4,095. Raise it toward the model's window.

The API docs list every error code.

Sources

Menu paths and field names are from SillyTavern's docs, read 2026-10-27: Chat Completions, Custom OpenAI-compatible endpoint, Common Settings and Reasoning. The request body, the default context and response sizes, and which samplers show for a Custom source are from SillyTavern's source on the release branch, read the same day. The Claude source's reverse proxy block, its Claude Model dropdown and the Prompt Post-Processing options are from SillyTavern's UI on the release branch, read 2026-11-02. For a chat client with a model allowlist and per-user keys, see the Open WebUI guide.

Questions

What are the best SillyTavern LLM settings?

On a hosted OpenAI-compatible API, start with Temperature 0.8 to 1.0, Top P 0.95, Frequency Penalty 0.1 to 0.3 and Presence Penalty 0 to 0.3, and set Context Size to the model's window from the Models page. Those four samplers are the ones a Custom source sends; tune them before anything else.

What does presence_penalty do in SillyTavern?

presence_penalty lowers the chance of any token that already appeared in the context, once, however often it appeared. It nudges the model toward new words and topics. frequency_penalty scales with how often a token appeared, so it targets repeated phrases. Both go to the model on a Custom source; 0 turns them off.

Do DRY settings work with an OpenAI-compatible API?

Not from the panel. DRY, XTC, Min P and Top K are Text Completion samplers, and SillyTavern does not send them for a Custom Chat Completion source. You can add a field under Additional Parameters > Include Body Parameters; the API forwards it, and whether the model applies it depends on the model.

What temperature should I use for roleplay?

Between 0.7 and 1.0 for most hosted models. Lower keeps a character consistent; higher gives more varied replies and more drift. If replies start to wander off the scene, lower temperature or Top P before adding penalties.

How much does a SillyTavern message cost?

It depends on the model and on how much history goes with each message. With 3,000 tokens in and 300 out, a message on Claude Sonnet 5.5 costs $0.0045 today and on GLM 5.3 $0.0013. A larger Context Size raises the input side once the chat fills it.

What is a SillyTavern proxy?

Usually a model backend: a URL and a key that answer in the OpenAI or Anthropic format. In SillyTavern that is the Custom (OpenAI-compatible) source, or the Claude source with a Proxy Server URL. The config.yaml requestProxy setting is something else, a network proxy for SillyTavern's own traffic, and so is the reverse proxy you put in front of SillyTavern for remote access.

Is there a free SillyTavern proxy?

Not here. RelaySwitch is prepaid from $10 by card or crypto, billed per token. Before the first top-up the console chat gives 12 free turns a day to try models, in the browser, not through SillyTavern.

Does SillyTavern need the /v1 at the end?

Yes. Use https://api.relayswitch.io/v1 as the Custom Endpoint (Base URL) or Proxy Server URL. SillyTavern adds /chat/completions (or /messages on the Claude source) after it, and a URL without /v1 or with the suffix already on it fails to connect.

Can I use the Claude source with a custom URL?

Yes. Set Chat Completion Source to Claude, put https://api.relayswitch.io/v1 in Proxy Server URL and your sk-relay- key in Proxy Password. The model comes from the Claude Model dropdown, which lists fixed ids; for a Claude id it does not list, use the Custom source and type the id.

Every model in one SillyTavern connection

Prepaid from $10 by card or crypto. No KYC, no subscription.

Models Get a key

Originally published at relayswitch.io/guides/sillytavern; the prices there are read from the live catalog.

JanitorAI proxy setup: fields, models, price per message

UplinkSwitch — Mon, 02 Nov 2026 06:17:18 +0000

Checked 2026-11-02

The exact Proxy URL, the model ids to type, what one 2K-in / 300-out turn costs on each, and the proxy errors by code.

A Janitor AI proxy is a Provider setting that sends each message of a chat to a chat completions URL you choose instead of to Janitor's own model. The reply comes from the model id you type, and you pay that API per token. The Proxy URL for RelaySwitch is https://api.relayswitch.io/v1/chat/completions: the full path, with no extra slash and no repeated segment.

If you came from a proxy that started charging or ended its free tier, only three fields change: the URL above, a sk-relay- key from the console, and a model id from the table below. Keys are prepaid from $10 by card or crypto, with no subscription and no KYC. Every reply bills its tokens from that balance and carries a signed receipt, so the price tables on this page are computed from the live catalog, never typed.

The proxy configuration

Open any chat whose creator allows proxies. Click the gear or menu icon at the top right, choose API Settings, pick Proxy under Provider, then + Add Configuration (older layouts show the same panel behind the using janitor button above the chat, with + New). Five fields:

text

1Config Name: RelaySwitch

2Model Name: glm-5.3-flash-uncensored

3Proxy URL: https://api.relayswitch.io/v1/chat/completions

4API Key: sk-relay-...

5Custom Prompt: (leave blank)

Config Name: anything; it is the label in the list of configurations.
Model Name: the id, pasted exactly. Janitor sends it as the model field, so a display name or a stray space gets a 404.
Proxy URL: https://api.relayswitch.io/v1/chat/completions. Janitor posts to the address as given, so a URL that stops at /v1 fails.
API Key: the key, with no space before or after it.
Custom Prompt: leave it blank. The character card and your persona already go with every message; a prompt here is billed as input on each turn.

Click Add, then Save Settings, then refresh the page. Open the configuration again and check it reads Active. Skip the Check Key/Model button: Janitor's own quickstart calls it unreliable and says to save and refresh instead, and a Network Error after clicking it means nothing. Janitor relays proxy requests from its own servers, so the first message is the real test.

There is no max tokens or temperature field on the proxy form. Those stay in Janitor's generation settings and travel with each request unchanged.

Which model to type

Any chat id on the Models page goes in Model Name. The 10 builds the catalog tags for uncensored or roleplay use come first, then the flagships on the same key. Prices are USD per million tokens; the last two columns price one 2K-in / 300-out turn and 100 of them. There is no free tier: a $10 top-up covers about 30,179 such turns on GLM-5.3 Flash Uncensored.

Model	Model Name	Input	Output	one 2K-in / 300-out turn	100 turns
GLM-5.3 Flash Uncensored	`glm-5.3-flash-uncensored`	$0.1086	$0.3802	$0.00033	$0.0331
GLM-5.3 Uncensored	`glm-5.3-uncensored`	$0.291	$0.873	$0.00084	$0.0844
Gemma 4 31B Fabled	`gemma-4-31b-fabled`	$0.35	$1.575	$0.0012	$0.1173
Gemma 4 31B Novelist	`gemma-4-31b-novelist`	$0.35	$1.575	$0.0012	$0.1173
Qwen3.8 27B Novelist	`qwen3.8-27b-novelist`	$0.35	$2.10	$0.0013	$0.133
Cydonia 24B v4.3	`cydonia-24b-v4.3`	$0.484	$0.60	$0.0011	$0.1148
Uncensored AI Roleplay	`uncensored-ai-roleplay`	$0.60	$2.34	$0.0019	$0.1902
DeepSeek V4.1 Flash Abliterated	`deepseek-v4.1-flash-abliterated`	$0.875	$1.75	$0.0023	$0.2275
DeepSeek V4 Pro Uncensored	`deepseek-v4-pro-uncensored`	$1.00	$2.00	$0.0026	$0.26
MythoMax L2 13B	`mythomax-l2-13b`	$0.404	$0.404	$0.00093	$0.0929
Claude Sonnet 4.6	`claude-sonnet-4-6`	$1.80	$9.00	$0.0063	$0.63
Claude Opus 5.5	`claude-opus-5-5`	$2.80	$10.00	$0.0086	$0.86
Claude Fable 5	`claude-fable-5`	$3.00	$15.00	$0.0105	$1.05
Gemini 3.1 Pro Preview	`gemini-3.1-pro-preview`	$1.20	$7.20	$0.0046	$0.456
DeepSeek V3.2	`deepseek-v3.2`	$1.08	$1.6801	$0.0027	$0.2664
Grok 4.6	`grok-4.6`	$0.60	$1.80	$0.0017	$0.174
GLM 5.3	`glm-5.3`	$0.3276	$1.0296	$0.00096	$0.0964
GPT 5	`gpt-5`	$2.50	$20.00	$0.011	$1.10

USD, read from the RelaySwitch catalog when this page was built. A turn here is 2,000 tokens in (card, persona and recent history) and 300 out.

A build that thinks before it answers spends part of the reply budget on thinking. If replies come back short or cut off on one, raise max tokens in Janitor's generation settings to 1,000 or more, or pick a build without a thinking step. The uncensored models page lists every tagged build with its daily check, and the roleplay chat runs a character card on the same key in the browser, so you can try a model before typing it into Janitor.

What a long chat costs

Janitor resends the card, your persona and as much history as fits with every message, so a turn late in a chat sends far more than 2,000 tokens. The same 300-token reply at three history sizes:

Tokens sent per turn	GLM-5.3 Flash Uncensored	Uncensored AI Roleplay	Claude Sonnet 4.6
4,000	$0.00055	$0.0031	$0.0099
8,000	$0.00098	$0.0055	$0.0171
16,000	$0.0019	$0.0103	$0.0315

USD per turn, live catalog rates. The input side is what grows; the output price only matters for the reply itself.

When a model reports cached input tokens, they bill at its cached rate, which lowers a long chat that resends the same history. The console's Usage page lists every request with its tokens and charge, and each one has a receipt at GET /v1/requests/:id/receipt you can check on the verify page. The pricing page has the billing rules.

Janitor AI proxy errors by code

Janitor shows the status code the API returned, with its own heading. What each one means here, and the fix:

Janitor shows	Cause	Fix
ERROR 400: Bad Request	The model refused a field in the request, or the Custom Prompt made the body invalid.	Clear Custom Prompt, reset generation settings to the defaults, send again.
ERROR 401: Invalid Credentials	The key is missing, mistyped, revoked, or has a space around it.	Copy the key again from the console, paste it into API Key, Save Settings, refresh.
ERROR 402: Insufficient Credits	The prepaid balance is empty, or the key reached the budget set on it.	Top up, or raise the budget on the key in the console. The next message goes through.
ERROR 403	The key is narrowed to a model list and Model Name is not on it (scope_denied).	Add the model to the key, or change Model Name to one on its list.
No endpoints found for model name (404)	model_not_found: the id is misspelled or not in the catalog. route_not_found: the Proxy URL stops at /v1 or has a typo in the path.	Paste the id from the Models page. Set Proxy URL to https://api.relayswitch.io/v1/chat/completions exactly.
ERROR 429: Rate Limited	The key hit its requests-per-minute or tokens-per-minute limit.	Wait a few seconds before the next message or swipe, or raise the limit on the key.
ERROR 500 / 502 / Provider returned error / Model Down	A host serving the model answered an error. The API hands the request to the next host for that model at the same price; a 503 means none answered.	Send again. If it repeats, check the model on the status page or switch Model Name.
Unknown prompt response from worker	Janitor-side: the page lost its saved proxy state.	Refresh the page, confirm the configuration reads Active, send again.

To separate a key problem from a Janitor setting, list the models the key can call from any terminal:

bash

1curl https://api.relayswitch.io/v1/models -H "Authorization: Bearer sk-relay-..."

A working key returns a list whose id values are the exact strings for Model Name. The API docs list every error code, and status shows each model's daily canary.

Sources

Menu path, field names, the full-URL rule and the save-and-refresh flow are from Janitor's help center (help.janitorai.com): the TL;DR Quickstart Proxy Instructions article and its proxy error guides, read 2026-11-02. Error headings are quoted as Janitor prints them. Prices and model ids come from the RelaySwitch catalog at render. The same key runs SillyTavern, Chub AI and RisuAI; only the URL shape differs, and each guide states it.

Questions

What's a proxy on Janitor AI?

A Provider option that routes each chat message to a chat completions URL you supply instead of Janitor's own model. The reply comes from the model id in the configuration and the provider bills you per token. Janitor relays the request from its servers, so the provider sees Janitor's address, not yours.

What proxy should I use for Janitor AI?

One that publishes its prices per token, gives you your own key with a budget, and runs the models you want. Here one 2K-in / 300-out turn on GLM-5.3 Flash Uncensored costs $0.00033 at today's rate, every reply carries a signed receipt, and a key can be narrowed to a model list with a USD budget so a typo or a long night cannot overspend.

Is Janitor AI better with a proxy?

It is different: you choose the model and pay per token instead of using the default. A flagship such as Claude Sonnet writes longer, more consistent prose; an uncensored build answers the card as written; a small roleplay build costs a fraction of either. The table above prices the same turn on each so you can decide by the number.

How do I allow a proxy on a Janitor AI character?

The character’s creator decides that when editing the character; the proxy panel cannot override it. If a chat says proxies are not allowed, pick another character, or change the setting on a character you made yourself.

What API should I use for Janitor AI?

Any OpenAI-compatible chat completions API works. Janitor needs the full endpoint, so for RelaySwitch the Proxy URL is https://api.relayswitch.io/v1/chat/completions, the key starts with sk-relay-, and Model Name is any chat id on the Models page.

Is there a free API key for Janitor AI?

Not here. Keys are prepaid from $10 with no subscription, and that first top-up covers about 30,179 turns of 2,000 tokens in and 300 out on GLM-5.3 Flash Uncensored. Before the first top-up the console chat runs 12 free turns a day, which is enough to try a model before pasting it into Janitor.

How do I set up my own API in Janitor AI?

Create a key in the console, open a chat, go to API Settings, choose Proxy, click + Add Configuration, and fill Config Name, Model Name, Proxy URL (https://api.relayswitch.io/v1/chat/completions) and API Key. Click Add, Save Settings, refresh, and confirm the configuration reads Active.

What do API settings do in Janitor AI?

They choose where each message goes. Under Provider, the default sends it to Janitor’s own model; Proxy sends it to the URL in your configuration with your key and model id. Generation settings such as max tokens and temperature live in a separate panel and go with the request either way.

How do I get the proxy URL?

It is fixed: https://api.relayswitch.io/v1/chat/completions. Nothing is generated per account; only the key is yours. Paste it with no trailing slash and no extra path.

What is the best AI proxy for Janitor AI?

Judge by cost per turn at the history size you actually chat at, by whether the model answers your cards without a jailbreak prompt riding along as paid input, and by whether a failed request is retried on another host instead of landing as an error. The long-chat table above shows the first; the uncensored builds answer the second; failover answers the third.

Is the proxy free to use?

No. Each reply bills its input and output tokens from a prepaid balance at the rate on the Models page; a failed request is not charged. Top-ups start at $10 by card or crypto.

Why does Janitor show proxy error 401?

The API Key field is wrong: a mistyped or revoked key, or a space before or after it. Copy the key again from the console, paste it, Save Settings and refresh.

Why does Janitor show proxy error 404?

One of two fields. A Model Name the catalog does not list gets model_not_found; a Proxy URL that stops at /v1 or misspells the path gets route_not_found. Use an id from the Models page and https://api.relayswitch.io/v1/chat/completions exactly.

Why does Janitor show proxy error 429?

The key hit its per-minute request or token limit, usually from fast swipes. Wait a few seconds and send again, or raise the limit on the key in the console. A 429 from a host behind the model is retried on another host before you see it.

One Proxy URL, every model

Prepaid from $10 by card or crypto. No KYC, no subscription.

Uncensored models Get a key

Originally published at relayswitch.io/guides/janitorai; the prices there are read from the live catalog.

Uncensored AI models 2026: October’s builds, priced

UplinkSwitch — Tue, 27 Oct 2026 05:30:00 +0000

From the Uplink Switch LLC family: this post first appeared on UncensoredLLM (uncensoredllm.ai). Prices quoted are the live catalog figures at publication; the original page re-reads them at render.

A dated log of every uncensored, abliterated and roleplay build that joined the catalog in October 2026, by the day it went on sale: what each is, 100 messages at the live price and today’s check.

Published October 26, 2026 · prices as of October 27, 2026

Every uncensored AI model on this key went on sale in October 2026, and this page is the log: 134 builds by the day each one was listed, from the first five on October 11, 2026 to the latest on October 25, 2026. Each row carries what the build is, its live price per million tokens, what 100 chat messages cost at that price and whether it answered today’s real-model check as the build it is named for.

23 abliterated, 29 uncensored, 67 roleplay, 2 security, 13 house builds. 127 of the 134 passed the check run at 2026-10-27 04:35 UTC. Prices are from the catalog of October 27, 2026 and are read again every five minutes; nothing on this page is typed. A message is 2,000 input tokens (a card or system prompt with recent history) and 500 output tokens (the reply). The family pages group the same builds by base model; this page groups them by date.

October 2026 at a glance

Listed on	Builds	Kinds	Lowest per 100 messages	Passed today
October 11, 2026	5	2 abliterated, 3 uncensored	GLM-5.3 Flash Uncensored, $0.0407	5 of 5
October 12, 2026	1	1 roleplay	Roleplay Uncensored, $0.09	1 of 1
October 19, 2026	9	2 abliterated, 7 uncensored	GLM-4.7 Flash Uncensored, $0.054	9 of 9
October 20, 2026	90	18 abliterated, 6 uncensored, 65 roleplay, 1 security	Manta Flash 1.0, $0.042	87 of 90
October 21, 2026	24	13 uncensored, 1 security, 10 house builds	GPT OSS 120B Uncensored, $0.0583	20 of 24
October 22, 2026	4	1 roleplay, 3 house builds	Uncensored AI Roleplay, $0.237	4 of 4
October 25, 2026	1	1 abliterated	DeepSeek V4.1 Flash Abliterated Fast, $0.525	1 of 1

Listing dates from the catalog’s own record (read 2026-10-26); prices and verdicts live. 2 builds have no dated record and are left out: Red Team (Flash) and Red Team (Pro).

Across the month, 100 messages run from $0.0407 on GLM-5.3 Flash Uncensored to $3.12 on Unrestricted AI. The widest window is 1,048,576 tokens on GLM-5.3 Flash Apex Abliterated; 59 of the 134 take image input.

October 11, 2026: 5 builds, 2 abliterated and 3 uncensored

Model	Build	In / out $/M	100 messages	Check
GLM-5.3 Flash Uncensored glm-5.3-flash-uncensored	Uncensored	$0.1086 / $0.3802	$0.0407	Steady since Oct 12
GLM-5.3 Uncensored glm-5.3-uncensored	Uncensored	$0.282 / $0.846	$0.0987	Steady since Oct 12
GLM-5.3 Flash Abliterated glm-5.3-flash-abliterated	Abliterated	$0.88 / $0.88	$0.22	Steady since Oct 27
DeepSeek V4 Pro Uncensored deepseek-v4-pro-uncensored	Uncensored	$1.00 / $2.00	$0.30	Steady since Oct 12
GLM-5.3 Abliterated glm-5.3-abliterated	Abliterated	$2.25 / $4.50	$0.675	Steady since Oct 12

USD per million tokens and per 100 messages of 2,000 input and 500 output tokens, from the catalog of 2026-10-27; the check column is today’s. 5 of 5 passed.

October 12, 2026: 1 build, 1 roleplay

Model	Build	In / out $/M	Context	100 messages	Check
Roleplay Uncensored roleplay-uncensored	Roleplay	$0.30 / $0.60		$0.09	Steady since Oct 12

USD per million tokens and per 100 messages of 2,000 input and 500 output tokens, from the catalog of 2026-10-27; the check column is today’s. 1 of 1 passed.

October 19, 2026: 9 builds, 2 abliterated and 7 uncensored

Model	Build	In / out $/M	Context	100 messages	Check
GLM-4.7 Flash Uncensored glm-4.7-flash-uncensored	Uncensored	$0.12 / $0.60		$0.054	Steady since Oct 22
Gemma 4 26B-A4B Uncensored gemma-4-26b-a4b-uncensored	Uncensored	$0.27 / $1.26	33K	$0.117	Steady since Oct 22
DeepSeek V4.1 Flash Abliterated deepseek-v4.1-flash-abliterated	Abliterated	$0.8488 / $1.6975		$0.2546	Steady since Oct 19
Qwen3.8 27B Uncensored qwen3.8-27b-uncensored	Uncensored	$1.25 / $3.75	33K	$0.4375	Steady since Oct 22
Chat Uncensored chat-uncensored	Uncensored	$1.50 / $3.00		$0.45	Steady since Oct 19
Gemma 31B Heretic Uncensored gemma-31b-heretic-uncensored	Abliterated	$1.50 / $8.25		$0.7125	Steady since Oct 22
QwQ 32B Uncensored qwq-32b-uncensored	Uncensored	$2.64 / $4.00	24K	$0.728	Steady since Oct 20
Code Pro Uncensored code-pro-uncensored	Uncensored	$4.50 / $9.00		$1.35	Steady since Oct 19
GLM-5.3 Flash Code Uncensored glm-5.3-flash-code-uncensored	Uncensored	$6.00 / $6.00		$1.50	Steady since Oct 22

USD per million tokens and per 100 messages of 2,000 input and 500 output tokens, from the catalog of 2026-10-27; the check column is today’s. 9 of 9 passed.

October 20, 2026: 90 builds, 18 abliterated, 6 uncensored, 65 roleplay and 1 security

The largest batch of the month. The tables below split it by kind, lowest price per 100 messages first; most of the roleplay rows are Gemma 4 and Qwen3.8 tunes trained on long-form character chat, which answer at once with no reasoning step.

Abliterated: 18

Model	Build	In / out $/M	Context	100 messages	Check
Gemma 4 26B-A4B ChimeraX Heretic gemma-4-26b-a4b-chimerax-heretic	Abliterated	$0.484 / $1.52	262K	$0.1728	Steady since Oct 23
GLM-5.3 Flash Apex Abliterated glm-5.3-flash-apex-abliterated	Abliterated	$0.60 / $2.00	1049K	$0.22	Steady since Oct 22
Gemma 4 31B Cognitive Unshackled gemma-4-31b-cognitive-unshackled	Abliterated	$1.071 / $1.071	262K	$0.2678	Steady since Oct 20
Qwen3.5 27B Blossom V6.4 Derestricted qwen3.5-27b-blossom-v6.4-derestricted	Abliterated	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Qwen3.5 27B BlueStar v3 Derestricted qwen3.5-27b-bluestar-v3-derestricted	Abliterated	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Qwen3.5 27B Claude 4.6 Opus Reasoning Distilled Derestricted qwen3.5-27b-claude-4.6-opus-distilled-derestricted	Abliterated	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Qwen3.5 27B Claude 4.6 Opus Reasoning Distilled Derestricted Lite qwen3.5-27b-claude-4.6-opus-distilled-derestricted-lite	Abliterated	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Qwen3.5 27B Queen Derestricted qwen3.5-27b-queen-derestricted	Abliterated	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Gemma 4 31B Gembrain Heretic gemma-4-31b-gembrain-heretic	Abliterated	$1.224 / $1.224	262K	$0.306	Steady since Oct 20
Gemma 4 31B SDFT Heretic RP gemma-4-31b-sdft-heretic-rp	Abliterated	$1.224 / $1.224	262K	$0.306	Steady since Oct 20
NeuralDaredevil 8B Abliterated neuraldaredevil-8b-abliterated	Abliterated	$1.76 / $1.76	8K	$0.44	Steady since Oct 24
GLM-4.6 Derestricted v5 glm-4.6-derestricted-v5	Abliterated	$1.40 / $5.25	131K	$0.5425	Not checked yet
Qwen3.8 27B Obliterated qwen3.8-27b-obliterated	Abliterated	$1.00 / $7.00	524K	$0.55	Steady since Oct 22
DeepSeek R1 Distill Llama 70B Abliterated deepseek-r1-distill-llama-70b-abliterated	Abliterated	$2.80 / $2.80	16K	$0.70	Steady since Oct 20
Llama 3 70B Abliterated v3.5 llama-3-70b-abliterated	Abliterated	$2.80 / $2.80	8K	$0.70	Not checked yet
Llama 3.3 70B Abliterated llama-3.3-70b-abliterated	Abliterated	$2.80 / $2.80	33K	$0.70	Steady since Oct 22
MiMo V2.6 Flash Abliterated mimo-v2.6-flash-abliterated	Abliterated	$2.00 / $6.00	1049K	$0.70	Steady since Oct 22
DeepSeek R1 Distill Qwen 32B Abliterated deepseek-r1-distill-qwen-32b-abliterated	Abliterated	$5.60 / $5.60	16K	$1.40	Steady since Oct 23

Uncensored: 6

Model	Build	In / out $/M	Context	100 messages	Check
Dolphin Mistral 24B Venice Edition dolphin-mistral-24b-venice	Uncensored	$0.50 / $2.25	33K	$0.2125	Steady since Oct 20
Amoral Gemma3 27B v2 amoral-gemma3-27b-v2	Uncensored	$1.05 / $1.05	33K	$0.2625	Steady since Oct 26
Gemma 4 31B K1 v5 gemma-4-31b-k1-v5	Uncensored	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Hermes 3 Llama 3.1 70B hermes-3-llama-3.1-70b	Uncensored	$1.632 / $1.632	66K	$0.408	Steady since Oct 20
Hermes 4 405B hermes-4-405b	Uncensored	$1.20 / $4.80	128K	$0.48	Steady since Oct 20
MiMo V2.6 Flash Uncensored mimo-v2.6-flash-uncensored	Uncensored	$2.00 / $6.00	1049K	$0.70	Steady since Oct 22

Roleplay: 65

Model	Build	In / out $/M	Context	100 messages	Check
Manta Flash 1.0 manta-flash-1.0	Roleplay	$0.07 / $0.56	16K	$0.042	Answered the check
Manta Mini 1.0 manta-mini-1.0	Roleplay	$0.07 / $0.56	8K	$0.042	Answered the check
Gemma 4 12B Starry Ocean gemma-4-12b-starry-ocean	Roleplay	$0.175 / $0.875	131K	$0.0788	Steady since Oct 20
Gemma 4 12B StationKeeper gemma-4-12b-station-keeper	Roleplay	$0.175 / $0.875	131K	$0.0788	Steady since Oct 20
Magidonia 24B v4.3 magidonia-24b-v4.3	Roleplay	$0.3535 / $0.4235	33K	$0.0919	Steady since Oct 26
MythoMax L2 13B mythomax-l2-13b	Roleplay	$0.404 / $0.404	4K	$0.101	Steady since Oct 20
Gemma 4 31B Split-Untied gemma-4-31b-split-untied	Roleplay	$0.35 / $1.05	262K	$0.1225	Steady since Oct 21
Cydonia 24B v4.3 cydonia-24b-v4.3	Roleplay	$0.484 / $0.60	131K	$0.1268	Steady since Oct 20
Manta Pro 1.0 manta-pro-1.0	Roleplay	$0.21 / $1.75	66K	$0.1295	Answered the check
Artemis v1.1 artemis-v1.1	Roleplay	$0.35 / $1.575	262K	$0.1488	Steady since Oct 21
Gemma 4 31B Fabled gemma-4-31b-fabled	Roleplay	$0.35 / $1.575	262K	$0.1488	Steady since Oct 20
Gemma 4 31B Gembrain gemma-4-31b-gembrain	Roleplay	$0.35 / $1.575	262K	$0.1488	Steady since Oct 20
Gemma 4 31B Gemsicle gemma-4-31b-gemsicle	Roleplay	$0.35 / $1.575	262K	$0.1488	Steady since Oct 20
Gemma 4 31B Isometry gemma-4-31b-isometry	Roleplay	$0.35 / $1.575	262K	$0.1488	Steady since Oct 20
Gemma 4 31B MeroMero v2 gemma-4-31b-meromero-v2	Roleplay	$0.35 / $1.575	262K	$0.1488	Steady since Oct 20
Gemma 4 31B Novelist gemma-4-31b-novelist	Roleplay	$0.35 / $1.575	262K	$0.1488	Steady since Oct 20
Gemma 4 31B Scarlet gemma-4-31b-scarlet	Roleplay	$0.35 / $1.575	262K	$0.1488	Steady since Oct 20
Gemma 4 26B-A4B ChimeraX gemma-4-26b-a4b-chimerax	Roleplay	$0.42 / $1.33	262K	$0.1505	Steady since Oct 20
Gemma 4 26B-A4B Luminous Mirror gemma-4-26b-a4b-luminous-mirror	Roleplay	$0.42 / $1.33	262K	$0.1505	Steady since Oct 20
Gemma 4 26B-A4B MeroMero gemma-4-26b-a4b-meromero	Roleplay	$0.42 / $1.33	262K	$0.1505	Steady since Oct 20
Gemma 4 26B-A4B Moonlight Dusk gemma-4-26b-a4b-moonlight-dusk	Roleplay	$0.42 / $1.33	262K	$0.1505	Steady since Oct 20
Gemma 4 26B-A4B Musica gemma-4-26b-a4b-musica	Roleplay	$0.42 / $1.33	262K	$0.1505	Steady since Oct 20
Gemma 4 26B-A4B Opus Distill gemma-4-26b-a4b-opus-distill	Roleplay	$0.42 / $1.33	262K	$0.1505	Steady since Oct 20
Gemma 4 26B-A4B Shadow Siren gemma-4-26b-a4b-shadow-siren	Roleplay	$0.42 / $1.33	262K	$0.1505	Steady since Oct 20
Gemma 4 31B DarkIdol gemma-4-31b-darkidol	Roleplay	$0.40 / $1.80	262K	$0.17	Steady since Oct 20
Gemma 4 26B-A4B DarkSoul gemma-4-26b-a4b-darksoul	Roleplay	$0.484 / $1.52	262K	$0.1728	Steady since Oct 20
Qwen3.8 27B Antislop qwen3.8-27b-antislop	Roleplay	$0.35 / $2.10	262K	$0.175	Steady since Oct 20
Qwen3.8 27B Fimi qwen3.8-27b-fimi	Roleplay	$0.35 / $2.10	262K	$0.175	Steady since Oct 20
Qwen3.8 27B Kimiko qwen3.8-27b-kimiko	Roleplay	$0.35 / $2.10	262K	$0.175	Steady since Oct 20
Qwen3.8 27B Kiwen qwen3.8-27b-kiwen	Roleplay	$0.35 / $2.10	262K	$0.175	Steady since Oct 20
Qwen3.8 27B Novelist qwen3.8-27b-novelist	Roleplay	$0.35 / $2.10	262K	$0.175	Steady since Oct 20
Qwen3.8 27B Synthia qwen3.8-27b-synthia	Roleplay	$0.35 / $2.10	262K	$0.175	Steady since Oct 20
Stheno 8B v3.2 stheno-8b-v3.2	Roleplay	$0.7035 / $0.7035	16K	$0.1759	Steady since Oct 21
Qwen3.8 27B DarkIdol qwen3.8-27b-darkidol	Roleplay	$0.40 / $2.40	262K	$0.20	Steady since Oct 20
Veiled Calla 12B veiled-calla-12b	Roleplay	$1.05 / $1.05	33K	$0.2625	Steady since Oct 24
Gemma 4 31B Agares v1 gemma-4-31b-agares-v1	Roleplay	$1.071 / $1.071	262K	$0.2678	Steady since Oct 20
Gemma 4 31B Animus V14.1 gemma-4-31b-animus-v14.1	Roleplay	$1.071 / $1.071	262K	$0.2678	Steady since Oct 20
Gemma 4 31B AssGuard gemma-4-31b-assguard	Roleplay	$1.071 / $1.071	262K	$0.2678	Steady since Oct 24
Gemma 4 31B Garnet V2 gemma-4-31b-garnet-v2	Roleplay	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Gemma 4 31B Gembrain X Core gemma-4-31b-gembrain-x-core	Roleplay	$1.071 / $1.071	262K	$0.2678	Steady since Oct 20
Gemma 4 31B Isometry RP gemma-4-31b-isometry-rp	Roleplay	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Gemma 4 31B Queen gemma-4-31b-queen	Roleplay	$1.071 / $1.071	262K	$0.2678	Steady since Oct 20
Gemma 4 31B StyleTune gemma-4-31b-styletune	Roleplay	$1.071 / $1.071	262K	$0.2678	Steady since Oct 21
Anubis 70B v1.1 anubis-70b-v1.1	Roleplay	$1.085 / $1.085	32K	$0.2713	Steady since Oct 20
Gemma 4 31B Dark Gemistry gemma-4-31b-dark-gemistry	Roleplay	$1.224 / $1.224	262K	$0.306	Steady since Oct 20
Euryale 70B v2.2 euryale-70b-v2.2	Roleplay	$1.224 / $1.428	20K	$0.3162	Steady since Oct 24
Cydonia 24B v4.1 cydonia-24b-v4.1	Roleplay	$1.20 / $2.00	131K	$0.34	Steady since Oct 20
Cu-Mai R1 70B cu-mai-r1-70b	Roleplay	$1.7255 / $1.7255	33K	$0.4314	Steady since Oct 20
Hanami 70B x1 hanami-70b-x1	Roleplay	$1.7255 / $1.7255	33K	$0.4314	Steady since Oct 20
Mag Mell R1 12B mag-mell-12b-r1	Roleplay	$1.7255 / $1.7255	16K	$0.4314	Steady since Oct 23
NemoMix Unleashed 12B nemomix-unleashed-12b	Roleplay	$1.7255 / $1.7255	33K	$0.4314	Steady since Oct 23
Starcannon Unleashed 12B v1.0 starcannon-unleashed-12b	Roleplay	$1.7255 / $1.7255	16K	$0.4314	Steady since Oct 24
Qwen3.8 27B Fable qwen3.8-27b-fable	Roleplay	$0.875 / $5.25	524K	$0.4375	Steady since Oct 20
Qwen3.8 27B Hemingway qwen3.8-27b-hemingway	Roleplay	$0.875 / $5.25	262K	$0.4375	Steady since Oct 20
Qwen3.8 27B Queen qwen3.8-27b-queen	Roleplay	$0.875 / $5.25	524K	$0.4375	Steady since Oct 20
Euryale 70B v2.3 euryale-70b-v2.3	Roleplay	$1.972 / $1.972	20K	$0.493	Steady since Oct 21
Magnum Diamond 24B magnum-diamond-24b	Roleplay	$1.972 / $1.972	33K	$0.493	Steady since Oct 24
UnslopNemo 12B v4.1 unslopnemo-12b-v4.1	Roleplay	$1.972 / $1.972	8K	$0.493	Steady since Oct 20
The Omega Directive 24B v2.0 omega-directive-24b-v2	Roleplay	$2.00 / $2.00	33K	$0.50	Steady since Oct 20
Skyfall 36B v2 skyfall-36b-v2	Roleplay	$1.925 / $2.80	33K	$0.525	Steady since Oct 20
ReMM SLERP L2 13B remm-slerp-l2-13b	Roleplay	$2.036 / $3.78	6K	$0.5962	Steady since Oct 20
Wayfarer Large 70B wayfarer-large-70b	Roleplay	$2.4535 / $2.4535	33K	$0.6134	Not checked yet
Aion RP Llama 3.1 8B aion-rp-llama-3.1-8b	Roleplay	$3.20 / $6.40	33K	$0.96	Steady since Oct 24
DansPE 1.3 24B dans-pe-1.3-24b	Roleplay	$2.908 / $11.632	33K	$1.1632	Steady since Oct 21
Magnum v4 72B magnum-v4-72b	Roleplay	$8.024 / $11.968	16K	$2.2032	Steady since Oct 20

Security: 1

Model	Build	In / out $/M	Context	100 messages	Check
CyberHeabsy cyberheabsy	Security	$5.25 / $15.75	1049K	$1.8375	Steady since Oct 24

USD per million tokens and per 100 messages of 2,000 input and 500 output tokens, from the catalog of 2026-10-27; the check column is today’s. 87 of 90 passed.

October 21, 2026: 24 builds, 13 uncensored, 1 security and 10 house builds

Model	Build	In / out $/M	Context	100 messages	Check
GPT OSS 120B Uncensored gpt-oss-120b-uncensored	Uncensored	$0.1095 / $0.728	131K	$0.0583	Steady since Oct 21
DeepSeek V4 Flash 0731 Uncensored deepseek-v4-flash-0731-uncensored	Uncensored	$0.4365 / $1.746	262K	$0.1746	Steady since Oct 21
GLM-5.3 Flash Cybersecurity glm-5.3-flash-cybersecurity	Security	$0.525 / $1.75	1049K	$0.1925	Steady since Oct 23
Hy3 Uncensored hy3-uncensored	Uncensored	$0.75 / $4.00	262K	$0.35	Steady since Oct 22
Laguna S 2.1 Uncensored laguna-s-2.1-uncensored	Uncensored	$0.75 / $4.00	262K	$0.35	Not checked yet
MiMo-V2.5 Uncensored mimo-v2.5-uncensored	Uncensored	$0.75 / $4.00	262K	$0.35	Steady since Oct 22
MiMo-V2.5-Pro Uncensored mimo-v2.5-pro-uncensored	Uncensored	$0.75 / $4.00	262K	$0.35	Not checked yet
MiniMax M3 Uncensored minimax-m3-uncensored	Uncensored	$0.75 / $4.00	262K	$0.35	Steady since Oct 22
GLM 5.2 Uncensored glm-5.2-uncensored	Uncensored	$1.095 / $4.68	262K	$0.453	Steady since Oct 22
Nemotron 3 Ultra 550B Uncensored nemotron-3-ultra-550b-a55b-uncensored	Uncensored	$1.125 / $6.60	262K	$0.555	Not checked yet
Kimi K2.6 Uncensored kimi-k2.6-uncensored	Uncensored	$1.17 / $7.34	262K	$0.601	Steady since Oct 22
GLM 4.7 Uncensored glm-4.7-uncensored	Uncensored	$1.527 / $7.27	131K	$0.6689	Steady since Oct 22
Inkling Uncensored inkling-uncensored	Uncensored	$2.13 / $12.14	262K	$1.033	Not checked yet
No Filter AI no-filter-ai	House build	$5.85 / $7.80	197K	$1.56	Steady since Oct 22
NSFW AI Chat nsfw-ai-chat	House build	$5.85 / $7.80	131K	$1.56	Steady since Oct 22
Uncensored AI Assistant uncensored-ai-assistant	House build	$5.85 / $7.80	197K	$1.56	Steady since Oct 22
Uncensored AI Companion uncensored-ai-companion	House build	$5.85 / $7.80	131K	$1.56	Steady since Oct 22
Uncensored AI Writer uncensored-ai-writer	House build	$5.85 / $7.80	131K	$1.56	Steady since Oct 22
Uncensored Chatbot uncensored-chatbot	House build	$5.85 / $7.80	197K	$1.56	Steady since Oct 22
Unfiltered AI unfiltered-ai	House build	$5.85 / $7.80	197K	$1.56	Steady since Oct 22
Kimi K3 Uncensored kimi-k3-uncensored	Uncensored	$5.745 / $28.68	262K	$2.583	Steady since Oct 22
Uncensored AI Chat uncensored-ai-chat	House build	$11.70 / $15.60	197K	$3.12	Steady since Oct 22
Uncensored Code Assistant uncensored-code-assistant	House build	$11.70 / $15.60	131K	$3.12	Steady since Oct 22
Unrestricted AI unrestricted-ai	House build	$11.70 / $15.60	131K	$3.12	Steady since Oct 22

USD per million tokens and per 100 messages of 2,000 input and 500 output tokens, from the catalog of 2026-10-27; the check column is today’s. 20 of 24 passed.

October 22, 2026: 4 builds, 1 roleplay and 3 house builds

Model	Build	In / out $/M	Context	100 messages	Check
Uncensored AI Roleplay uncensored-ai-roleplay	Roleplay	$0.60 / $2.34	128K	$0.237	Steady since Oct 22
Jailbreak-Free Assistant jailbreak-free-assistant	House build	$0.75 / $2.50	1049K	$0.275	Steady since Oct 22
Uncensored AI Novelist uncensored-ai-novelist	House build	$0.75 / $3.00	262K	$0.30	Steady since Oct 22
Uncensored Coding Agent uncensored-coding-agent	House build	$0.90 / $3.00	1049K	$0.33	Steady since Oct 22

USD per million tokens and per 100 messages of 2,000 input and 500 output tokens, from the catalog of 2026-10-27; the check column is today’s. 4 of 4 passed.

October 25, 2026: 1 build, 1 abliterated

Model	Build	In / out $/M	Context	100 messages	Check
DeepSeek V4.1 Flash Abliterated Fast deepseek-v4.1-flash-abliterated-fast	Abliterated	$1.75 / $3.50	1000K	$0.525	Steady since Oct 25

USD per million tokens and per 100 messages of 2,000 input and 500 output tokens, from the catalog of 2026-10-27; the check column is today’s. 1 of 1 passed.

Which to pick

One row per kind: the lowest-priced build of that kind that passed today’s check, or the lowest-priced one when none has a verdict yet. The kind decides what the price buys: an abliterated build has the refusal direction removed from its weights, an uncensored build was trained or served to answer, a roleplay tune was trained on character chat and is usually the smaller, cheaper model.

Kind	Pick	100 messages	Check	Listed on
Abliterated	Gemma 4 26B-A4B ChimeraX Heretic	$0.1728	Steady since Oct 23	October 20, 2026
Uncensored	GLM-5.3 Flash Uncensored	$0.0407	Steady since Oct 12	October 11, 2026
Roleplay	Manta Flash 1.0	$0.042	Answered the check	October 20, 2026
Security	GLM-5.3 Flash Cybersecurity	$0.1925	Steady since Oct 23	October 21, 2026
House build	Jailbreak-Free Assistant	$0.275	Steady since Oct 22	October 22, 2026

The DeepSeek and GLM-5.3 builds reason before they answer: give them 1,000 tokens of reply room, or send reasoning_effort “none”. The roleplay tunes need neither. Each model page lists the Context Size and reply length to set in SillyTavern and Janitor AI.

The same two sessions on every day

Two fixed shapes so the days can be read against each other. A chat month is 1,000 turns of 2,000 input and 500 output tokens. A coding day is 40 requests of 6,000 input and 1,200 output tokens. Each row gives the lowest and highest build listed that day.

Listed on	Chat month	Coding day
October 11, 2026	from $0.4074 on GLM-5.3 Flash Uncensored to $6.75 on GLM-5.3 Abliterated	from $0.0443 on GLM-5.3 Flash Uncensored to $0.756 on GLM-5.3 Abliterated
October 12, 2026	$0.90 on Roleplay Uncensored	$0.1008 on Roleplay Uncensored
October 19, 2026	from $0.54 on GLM-4.7 Flash Uncensored to $15.00 on GLM-5.3 Flash Code Uncensored	from $0.0576 on GLM-4.7 Flash Uncensored to $1.728 on GLM-5.3 Flash Code Uncensored
October 20, 2026	from $0.42 on Manta Flash 1.0 to $22.032 on Magnum v4 72B	from $0.0437 on Manta Flash 1.0 to $2.5002 on Magnum v4 72B
October 21, 2026	from $0.583 on GPT OSS 120B Uncensored to $31.20 on Unrestricted AI	from $0.0612 on GPT OSS 120B Uncensored to $3.5568 on Unrestricted AI
October 22, 2026	from $2.37 on Uncensored AI Roleplay to $3.30 on Uncensored Coding Agent	from $0.2563 on Uncensored AI Roleplay to $0.36 on Uncensored Coding Agent
October 25, 2026	$5.25 on DeepSeek V4.1 Flash Abliterated Fast	$0.588 on DeepSeek V4.1 Flash Abliterated Fast

Live from the catalog; input and output tokens only, no cache reads counted.

Call one of them

The id under each name is the model to send. Requests use the OpenAI chat completions format at https://api.uncensoredllm.ai/v1; SillyTavern takes the base URL ending in /v1, Janitor AI the full URL ending in /chat/completions. A request that fails before the model answers is not billed.

curl https://api.uncensoredllm.ai/v1/chat/completions \
  -H "Authorization: Bearer sk-ullm-live-…" \
  -H "Content-Type: application/json" \
  -d '{"model": "deepseek-v4.1-flash-abliterated-fast", "max_tokens": 1000, "messages": [{"role": "user", "content": "Hello"}]}'

The leaderboard ranks the same builds by today’s check and output price, and the uncensored LLM API page has the key, the SDKs and the clients.

Questions

What does uncensored mean for these builds?

Four things, by the build's kind. Abliterated: the refusal direction removed from the weights, nothing retrained. Uncensored: a fine-tune trained to answer, or the official weights served without a filter; where the id states no method, this page claims none. Roleplay: a fine-tune trained on long-form character chat. Security: a build carrying a security-analyst instruction. 23 abliterated, 29 uncensored, 67 roleplay, 2 security, 13 house builds on this page. See abliterated vs uncensored.

Were the new builds tested?

Each one gets the daily real-model check: one fixed prompt a day, with the verdict and the date it has held since. 127 of 134 passed the run at 2026-10-27 04:35 UTC. No refusal test is claimed for a build this page does not name one for; the check shows the model answering is the one named, not how uncensored it is. See the daily check.

What do the new uncensored models cost?

100 chat messages of 2,000 input and 500 output tokens cost from $0.0407 on GLM-5.3 Flash Uncensored to $3.12 on Unrestricted AI, from prepaid credit with no subscription. The model pages and the price index show the same figures. See the price index.

Which uncensored models are new this week?

The latest listing is October 25, 2026: DeepSeek V4.1 Flash Abliterated Fast. The day sections above are in date order, newest last.

Is there an abliterated Llama 4?

Not on this key. The Llama builds with the refusal removed are DeepSeek R1 Distill Llama 70B Abliterated, Llama 3 70B Abliterated v3.5 and Llama 3.3 70B Abliterated; the Llama 4 abliterated repositories on Hugging Face are Scout GGUFs for local use. See the Llama family.

Which of the new models take images?

59 of the 134, per the catalog's input field: Gemma 4 12B Starry Ocean, Gemma 4 12B StationKeeper, Artemis v1.1, Gemma 4 31B Fabled, Gemma 4 31B Gembrain, Gemma 4 31B Gemsicle, Gemma 4 31B Isometry and Gemma 4 31B MeroMero v2 and 51 more. Send images as image_url parts in the OpenAI chat format.

Can I download them instead?

The community builds are on Hugging Face under their own cards, which each model page names with its license; the official weights served without a filter and the house builds have no card to download. Calling any of them here needs no download and no GPU: a base URL, a key and the model id. See the models by family.

Get a key Prepaid from $10, no subscription.

Originally published at https://uncensoredllm.ai/blog/uncensored-models-october-2026.

Free Claude API: what exists in 2026, cheapest paid route

UplinkSwitch — Tue, 27 Oct 2026 05:24:29 +0000

From the Uplink Switch LLC family: this post first appeared on RelaySwitch (relayswitch.io). Prices quoted are the live catalog figures at publication; the original page re-reads them at render.

There is no free Claude API tier anywhere real. What Anthropic's subscriber credits are, what the proxies on GitHub do, and what every Claude model costs per million tokens here, from a $10 top-up by card or crypto.

There is no free Claude API. Anthropic's pricing page lists a per-token price for every model and no free tier; the threads that rank for the phrase are about subscriber credits, proxies and trial coupons. This post says what each of those is, then gives the number that matters: what every Claude model costs per million tokens on this key today, read from the live catalog, from a $10 top-up by card or crypto.

The floor, today

Claude Haiku 5.5 is $0.06 in and $0.15 out per million tokens: about $0.0002 for a chat turn of 2,000 tokens in and 500 out, or roughly 51,282 turns on a $10 top-up.

What "free Claude API" turns out to be

Anthropic's subscriber API credits. A monthly allowance attached to some paid consumer plans, spent through a Console organisation you link. It is a perk of a subscription, not a free tier, and it is capped.
Proxies on GitHub. Repositories that pool donated keys, or that put a Claude-shaped endpoint in front of an open-weight model. The first puts your prompts through someone else's account; the second is not Claude.
Trial credits. A few dollars on a new account at some gateways, gone in a day of real use. Useful for a test, not a plan.
Cheap API landings. Pages that say "free Claude" and mean pay-as-you-go from zero. That is what this site is, without the word free in front of it.

Every Claude model on the key, per million tokens

Model	Input	Cached	Output	Developer list in / out
Claude Haiku 5.5	$0.06 (40% under)	$0.01	$0.15 (70% under)	$0.10 / $0.50
Claude Sonnet 4 5 20250929	$0.50	$0.05	$2.4706	—
Claude Haiku 4.5	$0.60 (40% under)	$0.06	$3.00 (40% under)	$1.00 / $5.00
Claude Haiku 4.5	$0.60 (40% under)	$0.06	$3.00 (40% under)	$1.00 / $5.00
Claude Sonnet 5	$1.20 (40% under)	$0.12	$3.00 (70% under)	$2.00 / $10.00
Claude Sonnet 5.5	$1.20 (40% under)	$0.075	$3.00 (70% under)	$2.00 / $10.00
Claude Fable 5 1	$0.9341	$0.0212	$4.6706	—
Claude Opus 4.6	$1.80 (64% under)	$0.16	$7.50 (70% under)	$5.00 / $25.00
Claude Opus 4.7	$1.80 (64% under)	$0.16	$7.50 (70% under)	$5.00 / $25.00
Claude Opus 5	$1.80 (64% under)	$0.16	$7.50 (70% under)	$5.00 / $25.00
Claude Opus 5.5	$2.716 (32% under)	$0.1455	$9.70 (51% under)	$4.00 / $20.00
Claude Fable 5	$3.00 (70% under)	$0.30	$15.00 (70% under)	$10.00 / $50.00
Claude Fable 5.1	$6.00 (40% under)	$0.15	$30.00 (40% under)	$10.00 / $50.00
Claude Opus 4 5 20251101	$14.1176	$1.4118	$70.5882	—

Cheapest output first; USD per million tokens, read at render; Anthropic's list where the catalog has it. Thinking variants are on the Claude price hub.

What a $10 top-up buys

Model	Per chat turn	Turns per $10
Claude Haiku 5.5	$0.0002	51,282
Claude Sonnet 4 5 20250929	$0.0022	4,473
Claude Haiku 4.5	$0.0027	3,703
Claude Haiku 4.5	$0.0027	3,703
Claude Sonnet 5	$0.0039	2,564
Claude Sonnet 5.5	$0.0039	2,564

A turn is 2,000 tokens in and 500 out at today's price; rounded down.

The cheapest paid route, step by step

An email address opens the account. No identity documents, no phone number, no Anthropic account.
A top-up from $10 by card, Apple Pay, PayPal, or USDT, USDC, BTC, ETH, SOL. Nothing is added on top.
A key, with a monthly spend limit if you want one. The Claude ids answer both the Anthropic Messages format and the OpenAI chat format.

Base URL https://api.relayswitch.io/v1, your key (it starts with sk-relay-), the model id. Nothing else changes.

python

1from openai import OpenAI

3client = OpenAI(base_url="https://api.relayswitch.io/v1", api_key="sk-relay-...")

4reply = client.chat.completions.create(

5 model="claude-haiku-5-5",

6 messages=[{"role": "user", "content": "Say hello in five words."}],

7)

8print(reply.choices[0].message.content)

bash

1curl https://api.relayswitch.io/v1/chat/completions \

2 -H "Authorization: Bearer $RELAYSWITCH_KEY" \

3 -H "Content-Type: application/json" \

4 -d '{"model": "claude-haiku-5-5", "messages": [{"role": "user", "content": "Hello"}]}'

For Claude Code and the Anthropic SDK, set ANTHROPIC_BASE_URL to the base URL and ANTHROPIC_AUTH_TOKEN to the key; the Claude Code guide has the exact lines.

NSFW, roleplay and what Claude will not do

The Claude ids follow Anthropic's usage policy here exactly as they do at Anthropic. If the search that brought you here was about Janitor AI, SillyTavern or an unfiltered assistant, the uncensored builds on this key are the models made for that, on the same balance and the same base URL, and the roleplay client runs them in the browser.

Questions

Can I get the Claude API for free?

Not as a free tier. Anthropic has none: its API is prepaid per token, and the monthly API credit it gives some subscription plans is a perk of paying for the plan, spent through a Console organisation. The GitHub proxies that say "free Claude API" pool other people's keys or route to a different model. Here the floor is a $10 top-up, and Claude Haiku 5.5 costs about $0.0002 a chat turn at today's price.

Does the Claude API allow NSFW?

No. The Claude ids answer as Anthropic serves them, under Anthropic's usage policy, here as anywhere; a request that policy refuses comes back refused. For unfiltered chat and roleplay the uncensored builds on the same key are the models made for it, priced per token with the same balance.

Is there a free API for Claude Code on GitHub?

The repositories that rank for that are proxies: they pool keys, or point Claude Code at an open-weight model under a Claude-shaped endpoint. Claude Code itself runs on this key with two environment variables (ANTHROPIC_BASE_URL=https://api.relayswitch.io/v1, your key as ANTHROPIC_AUTH_TOKEN) and bills per token from the balance; the Claude Code guide has the exact lines.

Can I get any AI API for free?

Some providers run free tiers on small open-weight models with daily caps, and most hand new accounts a trial credit for a few days. For the Claude models specifically, no. What exists is a metered price and the choice of who bills you for it.

What is the cheapest paid way to call Claude?

On this key Claude Sonnet 5.5 is $1.20 in and $3.00 out per million tokens today, against Anthropic's list of $2.00 and $10.00; Claude Haiku 5.5 is cheaper still at $0.06 / $0.15. Prepaid from $10, by card or in USDT, USDC, BTC, ETH, SOL, no identity check, no subscription.

Originally published at https://relayswitch.io/blog/free-claude-api.

Best uncensored AI image generators 2026, ranked

UplinkSwitch — Tue, 27 Oct 2026 05:18:58 +0000

The best uncensored AI image generators of 2026, ranked by three stated tests: price per image from a published list, an API or a web app, and whether you must sign up. Our rows are priced live.

Published October 27, 2026

The best uncensored AI image generators of 2026 split three ways: a free web app, open weights on your own card, or an API that runs those weights with the safety checker off. This ranking states its test for each row and reads our own prices from the live catalog, so the numbers on this page are the numbers you would pay today.

The 2026 ranking at a glance

Every page for this search ranks itself first and states no test. This one ranks by three facts you can check on each tool’s own page: what one image costs from a published price list, whether there is an API or only an app, and whether you have to sign up. The other tools were read on October 27, 2026; our rows are read from the live catalog each time the page renders.

#	Generator	Pick	Type	From, per image	Sign-up
1	UncensoredLLM media API	Best uncensored image generator API	API	$0.0136	Email, prepaid from $10
2	Perchance	Best free, no sign-up	Web app	their pricing page	None
3	Venice	Best hosted app	Web app + API	their pricing page	Account for the paid tier
4	ComfyUI with FLUX or Stable Diffusion 3.5	Best local	Local	their pricing page	None
5	Mage.space	Best checkpoint choice in a browser	Web app	their pricing page	Account
6	Tensor.Art	Most community models	Web app	their pricing page	Account
7	NovelAI	Best for anime	Web app	their pricing page	Subscription

Our prices from the catalog of October 27, 2026; the other rows from each site on October 27, 2026. A tool’s price is never typed here: the link goes to the page that holds it.

Best uncensored AI image generator API: every model, priced live

11 uncensored image models answer on one key: 9 that draw from a prompt and 2 that edit an image you send. The safety checker is a plain parameter that defaults to off, and the price is per image, not per month. The range today is $0.0136 to $0.0938 per image, from the catalog of October 27, 2026.

Model	Draws or edits	From	Per
CyberRealistic XL Uncensored	Draws	$0.0136	image
WAI Illustrious SDXL Uncensored	Draws	$0.0136	image
Anime Image Uncensored	Draws	$0.03	image
Lustify SDXL Uncensored	Draws	$0.0313	image
Seedream 5.0 Lite Edit Uncensored	Edits	$0.0313	image
Qwen Image 2.1 Uncensored	Draws	$0.0375	image
Vision Image Uncensored	Draws	$0.04	image
Chroma Uncensored	Draws	$0.08	image
Qwen Edit Uncensored	Edits	$0.08	image
Qwen Image 3 Uncensored	Draws	$0.08	image
Vision Pro Image Uncensored	Draws	$0.0938	image

At the lowest row, 100 images cost $1.36 and the $10 minimum top-up buys 735 of them. The image API price guide works through what each setting does to the price.

One request, the OpenAI way. The id in the table is the id you send; any client that speaks the OpenAI images endpoint works with the base URL and the key changed.

curl https://api.uncensoredllm.ai/v1/images/generations \
  -H "Authorization: Bearer sk-ullm-live-…" \
  -H "Content-Type: application/json" \
  -d '{"model": "anime-image-uncensored", "prompt": "a lighthouse at dusk, oil on canvas", "size": "1024x1024", "n": 1}'

The answer is a URL on this API that stays valid for 30 days. The request body is not stored unless the key opts in to the 7-day replay, so the prompt is gone once the image is made. A key takes an email and a top-up from $10 in the console; there is no subscription.

2. Perchance: best free, no sign-up

A free browser generator with no login and no stated limit, which is why it owns the "no sign-up" searches. Settings are a text box and a style list; the model behind it is not named on the page.

Type: Web app. Sign-up: none. Price: on their own page, read October 27, 2026; re-check it before you decide, since plans move.

3. Venice: best hosted app

A private, uncensored chat and image app with a free tier and a paid plan, plus an API. Its own guide to unrestricted generators is the first organic result for that phrase, so it is the hosted app most people compare against.

Type: Web app + API. Sign-up: account for the paid tier. Price: on their own page, read October 27, 2026; re-check it before you decide, since plans move.

4. ComfyUI with FLUX or Stable Diffusion 3.5: best local

Open weights on your own graphics card: no filter at all, no per-image bill, prompts that never leave the machine. The price is the hardware (a 12 GB card for FLUX.1-dev at reduced precision) and the hours spent on node graphs and checkpoints.

Type: Local. Sign-up: none. Price: on their own page, read October 27, 2026; re-check it before you decide, since plans move.

5. Mage.space: best checkpoint choice in a browser

A hosted front end for open checkpoints with an NSFW switch on the paid plans; the Reddit list threads name it beside Perchance.

Type: Web app. Sign-up: account. Price: on their own page, read October 27, 2026; re-check it before you decide, since plans move.

6. Tensor.Art: most community models

Community checkpoints and LoRAs run in the browser on a credit system, with a free daily allowance; the AI Overview lists it for the model variety.

Type: Web app. Sign-up: account. Price: on their own page, read October 27, 2026; re-check it before you decide, since plans move.

7. NovelAI: best for anime

The anime generator with its own trained models, sold by subscription; the strongest pick when the style matters more than the price per image.

Type: Web app. Sign-up: subscription. Price: on their own page, read October 27, 2026; re-check it before you decide, since plans move.

Unrestricted AI image generator: what the word covers

Three different things hide behind “unrestricted”. A free web app with the filter set low by its owner, which can be turned back up any day. Open weights on your own card, where no filter exists at all. And an API that runs the same open weights with the safety checker off, which is what the first row is.

One rule holds on every uncensored model here, app or API: a prompt that puts a minor in the picture is refused before anything is charged. The acceptable-use page lists the rest. Adult content between adults, violence in fiction, and the styles the big generators decline are the normal case, and nothing rewrites the prompt on the way in.

For video the same split applies; the uncensored video generator API guide prices those models per second.

How the best uncensored AI image generators were ranked

Price per image, from a published list. Where a tool sells a subscription or credits, the cell links to its pricing page instead of a converted number.
API or app. A tool with an API can sit inside your own product; an app can only be used by hand.
Sign-up. None, an account, or a subscription, as the site asked on October 27, 2026.

Our own row carries its evidence: every price above is read from the catalog at render, and the daily real-model check publishes which chat models answered as themselves today. The other six entries are described in their own sites’ words; where a fact was not on the page, it is not here. For the open-weight route, the FLUX.1-dev model card states the licence and the hardware it expects.

Questions

Which AI is best for image generation for free?

For a free browser tool with no sign-up, Perchance was the no-login option the search results and the AI Overview named on 2026-10-27. For images made by code, there is no free tier here: a key is prepaid from $10 and each image is priced on its own line of the catalog. See every uncensored image and video model.

Is there an uncensored AI image generator with an API?

Yes. 11 uncensored image models answer on the OpenAI images endpoint with one key, priced per image, with the safety checker off by default. The curl above is a complete request. See the image API price guide.

Is there a realistic AI image generator that is free and uncensored?

The realistic open checkpoints (FLUX and Stable Diffusion 3.5 derivatives) are free to run on your own graphics card through ComfyUI. Hosted, the same families cost per image; the model pages here list which checkpoint each build is.

Is there an unrestricted AI image generator app?

Venice, Mage.space and Tensor.Art run as web apps and work on a phone in the browser; none of them is an API. An API model can be used from any app that lets you set a base URL and a key.

Is there an unrestricted AI video generator too?

Yes. The uncensored video models on this key are priced per second of output and take the same request shape as the image models, with a submit-and-poll endpoint for long renders. See uncensored video generator API prices.

What does an uncensored AI image generator refuse?

Any prompt that puts a minor in the picture, on every model here, before the request is priced. The acceptable-use page lists the other limits; adult content between adults is not one of them. See the acceptable-use page.

Get a key Prepaid from $10, no subscription.

Originally published at https://uncensoredllm.ai/blog/best-uncensored-ai-image-generators-2026.

GPT-5.5 API key: pricing and how to get one, no KYC

UplinkSwitch — Tue, 27 Oct 2026 05:13:28 +0000

From the Uplink Switch LLC family: this post first appeared on RelaySwitch (relayswitch.io). Prices quoted are the live catalog figures at publication; the original page re-reads them at render.

GPT-5.5's price here next to OpenAI's list, read live, then the three-minute path to a key: an email, a top-up by card or crypto, one base URL. No identity documents, no subscription.

A GPT-5.5 API key costs what OpenAI charges per token, plus whatever the place you got the key from adds. OpenAI's own models page lists the price, and the gateways that rank for this phrase quote it with their fee on top. The table below is the price here, read from the live catalog as this page renders, next to OpenAI's list. After it: how to get the key in three minutes with an email and a top-up, no identity documents.

GPT-5.5 here today

$1.80 in, $9.00 out per million tokens, against OpenAI's list of $5.00 and $30.00 (64% under on input). Nothing added on the top-up.

GPT-5.5 API pricing per million tokens

Model	Input	Cached	Output	Developer list in / out
GPT-5.5	$1.80 (64% under)	$0.16	$9.00 (70% under)	$5.00 / $30.00
GPT-6 Astra	$3.00 (70% under)	$0.30	$15.00 (70% under)	$10.00 / $50.00
GPT-6.1 Sol	$0.60 (70% under)	$0.06	$3.00 (70% under)	$2.00 / $10.00
GPT-5.6 Sol	$1.20 (70% under)	$0.12	$3.00 (85% under)	$4.00 / $20.00
GPT-6 Luna	$0.0625 (37% under)	$0.0125	$0.20 (60% under)	$0.10 / $0.50

USD per million tokens, read at render; OpenAI's list where the catalog has it. The GPT price hub has every GPT id.

Two worked shapes, at today's price for GPT-5.5 and its neighbours:

Model	A 20k-token document, 1k out	Same at list
GPT-5.5	$0.0450	$0.1300
GPT-6 Astra	$0.0750	$0.2500
GPT-6.1 Sol	$0.0150	$0.0500
GPT-5.6 Sol	$0.0270	$0.1000
GPT-6 Luna	$0.0014	$0.0025

20,000 tokens in and 1,000 out, uncached. Cached input is cheaper where the model supports it.

How to get a GPT-5.5 API key without KYC

Open an account with an email address. No identity documents, no phone number, no OpenAI account.
Top up from $10: card, Apple Pay, PayPal, or USDT, USDC, BTC, ETH, SOL. The balance is USD and does not expire.
Create a key in the console and set a monthly spend limit if you want one.
Point the SDK at the base URL and use gpt-5.5 as the model id.

Base URL https://api.relayswitch.io/v1, your key (it starts with sk-relay-), the model id. Nothing else changes.

python

1from openai import OpenAI

3client = OpenAI(base_url="https://api.relayswitch.io/v1", api_key="sk-relay-...")

4reply = client.chat.completions.create(

5 model="gpt-5.5",

6 messages=[{"role": "user", "content": "Say hello in five words."}],

7)

8print(reply.choices[0].message.content)

bash

1curl https://api.relayswitch.io/v1/chat/completions \

2 -H "Authorization: Bearer $RELAYSWITCH_KEY" \

3 -H "Content-Type: application/json" \

4 -d '{"model": "gpt-5.5", "messages": [{"role": "user", "content": "Hello"}]}'

What the key does and does not change

The request and response shape is OpenAI's: messages, tools, response_format, streaming, the usage block. What changes is the base URL and who bills you. The model answers under OpenAI's usage policy here as anywhere.

One balance pays for every model on the key, so a GPT-5.5 app can fall back to a Claude, Gemini or DeepSeek id by changing one string. The OpenRouter alternatives page compares this against the other places to get one key for many models.

"GPT-5.5 free": what the search actually turns up

Trial credits on new accounts, proxies that pool keys, and tutorials that put a browser SDK in front of the model with the viewer paying. None is a free GPT-5.5 API key, and the proxies put your prompts through someone else's account. A $10 top-up here is the honest floor; the per-request figures above say how far it goes.

Questions

Is there a free GPT-5.5 API key?

No. OpenAI bills GPT-5.5 per token, and so does every gateway that resells it. The videos that promise a free key route you to a trial credit or a proxy. Here the smallest way in is a $10 top-up, with no subscription and no expiry on the balance.

How much does the GPT-5.5 API cost?

On this key $1.80 in and $9.00 out per million tokens today, against OpenAI's list of $5.00 and $30.00; cached input $0.16. The figures are read from the catalog at render and are what you are billed.

Do I need an OpenAI account to get a GPT-5.5 API key?

Not here. The account is an email address and a prepaid top-up; the GPT ids answer the OpenAI chat format at https://api.relayswitch.io/v1, so the OpenAI SDK works unchanged with the base URL and key swapped.

Does GPT-5.5 here support tools, JSON mode and streaming?

Yes: the request and response shapes are the OpenAI ones, including tool calls, response_format and streamed chunks. The docs list the fields the gateway passes through.

Can I pay for the GPT-5.5 API with crypto?

Yes: USDT, USDC, BTC, ETH, SOL to the same prepaid balance a card produces, with nothing added on the top-up and no identity check.

Originally published at https://relayswitch.io/blog/gpt-5-5-api-key.

Character AI alternatives: 12 apps and front-ends

UplinkSwitch — Tue, 27 Oct 2026 05:07:57 +0000

Pick a character AI alternative by route: eight companion apps on a subscription, or a front-end you control with an uncensored model billed per message.

Published October 24, 2026 · prices as of October 27, 2026

A character AI alternative is a choice between two routes more than a choice between two apps. One is a hosted companion app that picks the model, sets the content rules and sells a monthly plan. The other is a front-end you control, connected to a model you choose on a key that bills per message. It is the same question as any AI with no restrictions: who decides what the model will refuse.

This guide ranks twelve options across both routes, with every competitor price dated and linked to its source. Then it covers what an alternative is, the four types, how to switch without losing your characters, and a live table of what 100 messages cost on four uncensored models. Pick an app if you want voice, a phone app and nothing to configure. Pick the second route if you want to choose the model, keep your cards and pay only for the messages you send.

Character AI alternatives at a glance

Twelve options in two routes. The price column is what each provider's own page said on October 24, 2026; a dash means the page did not state it, and the entry below links to the page instead.

#	Option	Route	Price basis (October 24, 2026)	Pick it for
1	SpicyChat	Hosted app	$5.95, $19.95 or $29.95 a month; core features free	Persona slots and memory that grow by tier
2	Kindroid	Hosted app	Subscription, price shown at checkout; 50-message preview	One companion with long-term memory and calls
3	Nomi	Hosted app	Free to start, no card	Short, medium and long-term memory
4	Janitor AI	Hosted app	-	Card-based chat on its own model or one you pick
5	Replika	Hosted app	-	A single long-running companion on a phone
6	Chai	Hosted app	-	A mobile app with a large active user base
7	Talefy	Hosted app	-	Interactive stories rather than one-on-one chat
8	CrushOn	Hosted app	-	Character chat from a public library
9	SillyTavern	Your front-end	Free, open source; you supply the model	Full control of cards, prompts and model
10	Chub	Your front-end	Its own Mars plan $20 a month, or your own API	Character cards with a choice of model
11	Janitor AI on a model you pick	Your front-end	The site, plus what your key costs	Keeping Janitor's interface with a model you pick
12	UncensoredLLM key	Your model	Per message, prepaid from $10.00; live table below	An uncensored model inside 9, 10 or 11

How we judged them

Most switches turn on what you can check before you pay. That means the price, the message limit and who decides what the model will refuse.

Inside each route, the order is how much of the price, the limits and the memory a reader can confirm on the provider's own public pages, read on October 24, 2026. Providers that publish more sit higher. It is not a score for writing quality, and an app lower down may suit you better.

For our own row the evidence is public too. Every model's per-token price is read from the live catalog when this page renders, and the daily real-model check sends one fixed prompt to every model each day and records whether the answer still matches. On October 27, 2026 it found 308 of 323 models stable.

Character AI alternatives for roleplay on a subscription

Route one is the closest to Character.AI itself. You open an app, pick or build a character and chat. The app runs the model, sets the content rules and bills a monthly plan. Nothing to configure, and most include extras a raw model does not have, such as voice, images or a phone app.

1. SpicyChat

SpicyChat publishes its tiers and what each one adds. Its pricing FAQ lists three plans, Get A Taste at $5.95, True Supporter at $19.95 and I'm All In at $29.95 a month, effective September 8, 2026 (SpicyChat pricing FAQ, read October 24, 2026). Its docs say the core features stay free for everyone.

The tiers differ in memory and persona slots. Context memory is 4K, 8K and up to 32K tokens by tier, and personas go from 10 to 50 to 100 (premium features page, read October 24, 2026). Pick it if you want to see exactly what each step up buys. The trade is that the longest memory and the full model list sit on the top tier.

2. Kindroid

Kindroid is a companion app built around one character, a Kin, with its own backstory, voice and long-term memory, and it can take voice and video calls (kindroid.ai, read October 24, 2026). New accounts get a one-time preview of 50 messages, 15 in some regions, and subscribers get unlimited access to its newest flagship models (Kindroid plans, read October 24, 2026).

It sells monthly, quarterly and yearly terms and shows the exact price for your country at checkout, so there is no single dollar figure to quote. Pick it for one companion that recalls past conversations. The account turns read-only when the preview or subscription ends.

3. Nomi

Nomi describes itself as a companion with short, medium and long-term memory, and its home page says it is free to start chatting with no credit card (nomi.ai, read October 24, 2026). The home page does not list a subscription price.

Pick it to try a memory-first companion without entering a card. For the paid plan, check the price in the app before you subscribe.

4. Janitor AI

Janitor AI hosts community character cards and runs them on its own model, JLLM. Its help center puts JLLM's memory at about 8,000 to 9,000 tokens (Janitor AI help center, read October 24, 2026). We could not read a price on janitorai.com that day, so see their site for current terms.

It can also run the same cards on a model of your choice, which puts it in route two as well, entry 11 below.

5. Replika

Replika's home page says it built the world's first AI companion ten years ago, counts 42,160,934 users worldwide and lists apps for iOS and Android (replika.com, read October 24, 2026). The home page does not state a price; see their site or app store listing for the current plans.

Pick it if you want one companion that knows you, on a phone, rather than a library of characters.

6. Chai

Chai's site reports 10M+ active users (chai-research.com, read October 24, 2026). We found no price or message limit on that page; see the app for the current plans. Pick it for a mobile-first app with a large active user base.

7. Talefy

Talefy is built for stories more than chat: its FAQ describes reading interactive AI stories or generating your own plots (talefy.ai, read October 24, 2026). The FAQ points readers to the platform for current pricing. Pick it if your Character.AI use was mostly adventure and story scenes.

8. CrushOn

CrushOn runs character chat from a public library and ranks on the first page for this search with its own character chat page. Its site did not load for us on October 24, 2026, so we quote no price or limit; see crushon.ai for the current plans.

A front-end you control, a model you choose

Route two splits the app in two. The front-end holds your characters, personas and chat history. The model comes from an API key you choose, and you pay per message. Change the model and the characters stay; change the front-end and the key still works. The base URL, key and model id go in the same three fields in each supported client.

9. SillyTavern

SillyTavern is a locally installed interface, free and open source under the AGPL-3.0 licence. Its own docs are plain about the catch: it is only an interface, so you need a model backend for it to answer. It accepts any OpenAI-compatible API among others (SillyTavern documentation, read October 24, 2026).

Pick it for the most control: cards, lorebooks, prompt order and sampler settings all live in your own install.

10. Chub

Chub hosts character cards and connects them to several model sources. Its docs list OpenAI, Anthropic, Google, NovelAI, Kobold/Ooba and its own Mars models, with OpenAI-format connections by API key. A Mars subscription is $20 a month with unlimited messaging (Chub docs, read October 24, 2026).

Pick it if you want a hosted card library and the option to switch between a flat monthly plan and your own key.

11. Janitor AI on a model you pick

The same Janitor AI site, pointed at a model you pay for. Your cards and chats stay where they are; only the model answering them changes. Janitor's own help center carries guides for this next to its JLLM guides (user guides, read October 24, 2026). Pick it if you like Janitor's interface and want a model with a longer memory or no refusals.

12. An uncensored model on an UncensoredLLM key

This is where we fit, as the model half of route two. One prepaid key reaches every abliterated, uncensored and roleplay build on the uncensored model list, in the OpenAI-compatible format that SillyTavern, Chub and Janitor AI accept. The balance tops up from $10.00, there is no subscription, and request bodies are not stored unless a key opts in to a seven-day replay.

Abliterated builds have had the refusal behaviour removed from the weights, so there is no filter between the front-end and the answer. Some hosted apps filter adult content; these models do not refuse. The front-end's own rules still apply to what it lets you send.

Apps like Character AI with unlimited messages

Two of the options above state unlimited use on a flat plan, as of October 24, 2026. Kindroid subscribers get unlimited access to its newest flagship models. Chub's Mars plan is $20 a month with unlimited messaging. Both make sense if you chat every day and want a fixed monthly number.

A per-message key has no message cap at all. The balance is the limit, and each key can carry a spend limit set in the console, so a long weekend campaign stops at a number you chose. Your monthly message count times the per-message figure in the cost table further down is the whole bill.

Character AI alternatives for free

Free options exist on both routes, each with a limit. As of October 24, 2026:

SillyTavern is free and open source. It still needs a model to answer, which is where the cost sits.
Nomi is free to start chatting, with no card asked.
Kindroid gives new accounts a 50-message preview, 15 in some regions.
SpicyChat keeps its core features free for everyone.

What is a character AI alternative?

A character AI alternative is any service where you chat with a persistent character the way you would on Character.AI, without Character.AI. Every one of them has two parts. The character is a block of text: a name, a description, a scenario, a first message and a few lines of example dialogue. The model is what reads that text plus the chat so far and writes the next reply.

A hosted app ships both parts together and decides which model you get. A front-end you control keeps the character and lets you choose the model separately. That split is why the list above has two routes, and it is where most of the differences in price, memory and refusals come from.

Searches often add a qualifier, as in "character ai alternative no filter" or "c.ai alternatives no filter". The filter lives in two places: the app's content rules and the model's own refusals. A hosted app controls both. On the second route the front-end's rules still apply, but the refusals come from whichever model you connect, so an abliterated build removes that half.

Why people switch

People switch for a handful of reasons. A story gets cut off by a refusal halfway through a scene. A character forgets what happened twenty messages ago because the memory window is small. A daily cap or a queue slows a long session. Or the subscription price rises; SpicyChat's middle tier went from $14.95 to $19.95 on September 8, 2026, per its own pricing FAQ (read October 24, 2026).

Continuity is the newer reason to look for a Character.AI alternative. On October 12, 2026 Character.AI's blog said the company had initiated a voluntary Chapter 11 bankruptcy process to pursue a court-supervised sale, and that the platform remains up with subscriptions active for now (read October 24, 2026). Nothing on this page depends on how that ends. It does make the case for keeping your characters somewhere you control.

Four types of alternative

Hosted companion apps

Entries 1 to 8 above. The app holds your characters, runs its own model and bills a plan. Xoul AI is another in this group: its home page describes creating and chatting with AI characters, scenarios and worlds (xoul.ai, read October 24, 2026). Our Xoul AI guide puts an uncensored model in it. A scenario you write elsewhere, for example with the AI scenario generator, pastes into any of them as a character's opening.

Front-ends you run yourself

Software you install or open yourself, so the characters stay with you. SillyTavern is entry 9 above. KoboldAI Lite is a single static HTML file that opens in a browser, under the AGPL-3.0 licence (its repository, read October 24, 2026). Our KoboldAI Lite guide puts an uncensored model in it. RisuAI is a cross-platform chat app under GPL-3.0 (its repository, read October 24, 2026). Our RisuAI guide puts an uncensored model in it. HammerAI's desktop app can run models on your own machine through Ollama as well as cloud models (hammerai.com, read October 24, 2026). Our HammerAI guide puts an uncensored model in it.

Front-ends that take your own API key

Interfaces that hold your characters and send each turn to a model account you choose. Chub and Janitor AI work this way alongside their own models. Agnai describes itself as roleplay chat using your favourite AI services, under AGPL-3.0 (its repository, read October 24, 2026). Our Agnai guide puts an uncensored model in it. Novelcrafter is a writing platform that connects to an AI platform of your choice (novelcrafter.com, read October 24, 2026), which suits long-form story work more than chat. Our Novelcrafter guide puts an uncensored model in it.

Model routes

The model half of route two. You can run a model locally if you have the GPU for it, or call one by API from any device. On this site the API route is the list of uncensored models: abliterated, uncensored and roleplay builds on one key. The roleplay tunes sit with their families on the uncensored Gemma builds and Qwen builds pages.

How to switch without losing your characters

Copy every character out first. Save the name, description, scenario, first message and example dialogue of each character you care about as plain text. Front-ends you control store a character as a card you can move between them; keep a copy outside any single app.
Pick the route. Use the table at the top: an app if you want voice, a phone app and no setup; a front-end and a key if you want to choose the model and pay per message.
Pick a model. A roleplay tune for long in-character chat, a general uncensored build for mixed use. Check that its context window holds your card plus the history you want it to remember, and check its daily-check verdict.
Budget per message. Input tokens times the input rate plus output tokens times the output rate is one message. The cost table below does this at render for four builds; multiply by the messages you send in a month and compare with a plan price.
Cap the spend. Start with the smallest top-up and set a spend limit on the key, so a long session cannot run past a number you chose.
Restart the scene. Paste the old scenario and first message into the new home, or write a fresh opening and let the first few replies settle the character's voice.

Best practices

Keep the character description tight. A front-end resends it with every turn, so each extra paragraph is paid for on every message.
Set the context size to what the story needs, not the model's maximum. Summarize old turns into a paragraph once a chat runs long.
Try the same card on two or three models before you commit. Voice and pacing change from one model to the next.
Keep one model for one long campaign. Switching mid-story changes how the character talks.
Back up your cards and your best chats outside the app they were written in.

Common mistakes

Leaving the context setting at its maximum. The input side of the bill then grows with every turn of a long chat.
Comparing a monthly plan with a per-token rate. Convert both to a price per message at your own message size first.
Reading "unlimited" as covering everything. Kindroid's plans page words it as unlimited access to its newest flagship models (read October 24, 2026).
Keeping characters in one app only. If the app changes its rules, its price or its owner, the characters go with it.
Using a shared key from a stranger. Chub's docs recommend official API keys over third-party relays for security reasons (read October 24, 2026). Your own key, with your own spend limit, is the safe version.

Tools: message costs, models and scenarios

What 100 messages cost on an uncensored model

Four chat builds suited to character chat, read from the live catalog as this page renders. Rates are per million tokens.

Model	Build	Input $/M	Output $/M
Roleplay Uncensored roleplay-uncensored	Roleplay	$0.30	$0.60
Cydonia 24B v4.3 cydonia-24b-v4.3	Roleplay	$0.484	$0.60
Euryale 70B v2.2 euryale-70b-v2.2	Roleplay	$1.224	$1.428
GLM-5.3 Uncensored glm-5.3-uncensored	Uncensored	$0.282	$0.846

Prices per million tokens, read from the live catalog on 2026-10-27.

One message here is 1,500 input tokens (the character card plus the recent history a front-end resends every turn) and 250 output tokens back, about three or four paragraphs. The figures below are computed from the rates above at render time.

Model	Per message	100 messages	1,000 messages	Daily check
Roleplay Uncensored	$0.0006	$0.06	$0.60	Steady since Oct 12
Cydonia 24B v4.3	$0.0009	$0.09	$0.88	Steady since Oct 20
Euryale 70B v2.2	$0.0022	$0.22	$2.19	Steady since Oct 24
GLM-5.3 Uncensored	$0.0006	$0.06	$0.63	Steady since Oct 12

At this message size, 100 messages cost between $0.06 and $0.22 on these four builds, priced on 2026-10-27. The $19.95 of SpicyChat's middle tier on October 24, 2026 would pay for about 33,250 messages on Roleplay Uncensored. The tier also includes features a model key does not, such as conversation images.

Input decides the bill once a chat is long. A front-end resends the history every turn, so a lower context setting or a summary of old turns cuts the per-message figure. The per-token prices page has every model's rate and the bonus credit for larger top-ups.

The model list

Every uncensored build with its live input and output rate, context window and daily-check verdict is on the Models page. Roleplay tunes are labelled as such. Bonus credit tiers for larger top-ups are on Pricing, and the check itself, model by model, is on the proof page.

The scenario generator

If what you miss most is the opening of a scene, the roleplay scenario generator writes a scenario, characters and a first message with an uncensored model, two free per browser, ready to paste into any of the front-ends above.

Questions

Is there a character AI alternative with no filter?

Yes, on the second route. Hosted apps set their own content rules, and some filter adult content. An abliterated or uncensored model on your own key has had its refusals removed, so it answers what the front-end sends. Chub's docs describe Chub as uncensored, and its Mars models are its own uncensored suite (read October 24, 2026). See every uncensored model, priced.

Is Character.AI or Janitor AI better?

They suit different people. Character.AI is a polished hosted app on its own models; its blog said on October 12, 2026 that it had started a voluntary Chapter 11 process toward a court-supervised sale, with the platform still running and subscriptions active. Janitor AI runs community cards on its own model, JLLM, with about 8,000 to 9,000 tokens of memory per its help center, and it can also run cards on a model you pay for (both read October 24, 2026).

Which Character AI alternatives support personas?

SpicyChat ties persona slots to its tiers: up to 10, 50 or 100 (read October 24, 2026). Janitor AI's help center has a guide to writing a persona. In a front-end you control, such as SillyTavern or Chub, the persona is part of what you send to the model, so any model on your key uses it.

Are there Character AI alternatives on iOS?

Replika lists apps for iOS and Android on its home page (read October 24, 2026). On the second route, Janitor AI and Chub are websites that work in a phone browser, and any iOS client that accepts a custom OpenAI-compatible endpoint can use the same key. See the settings for each client.

Which apps like Character AI have unlimited messages?

Kindroid subscribers get unlimited access to its newest flagship models, and Chub's Mars plan is $20 a month with unlimited messaging (both read October 24, 2026). A per-message key has no message cap; the prepaid balance and an optional spend limit per key are the only limits.

What are the best Character AI alternatives for free?

SillyTavern is free and open source but needs a model behind it. Nomi is free to start with no card, Kindroid gives a 50-message preview, and SpicyChat keeps its core features free (all read October 24, 2026).

Get a key Prepaid from $10, no subscription.

Originally published at https://uncensoredllm.ai/blog/character-ai-alternatives.

Grok free tier API: gone, and what to use instead

UplinkSwitch — Tue, 27 Oct 2026 05:02:26 +0000

From the Uplink Switch LLC family: this post first appeared on RelaySwitch (relayswitch.io). Prices quoted are the live catalog figures at publication; the original page re-reads them at render.

xAI's free API tier ended in 2025. What the Grok API costs at xAI's list today, what the same models cost on this key per million tokens, and how to get a key with no monthly minimum, by card or crypto.

The Grok free tier API is gone. xAI opened its API in November 2024 with a monthly credit for every account, ended that allowance in 2025, and now publishes per-million-token prices with a free playground beside them. What people searching for the free tier want is a way to call Grok 4.6 or Grok 4.7 cheaply, without a monthly commitment and without opening another developer account. This post gives the straight answer and the prices, read from the live catalog as the page renders.

Grok 4.7 here today

$0.60 in, $1.80 out per million tokens, against xAI's list of $2.00 and $6.00. A $10 top-up buys about 4,761 chat turns of 2,000 tokens in and 500 out.

What happened to the Grok free tier API

xAI's API documentation describes a metered API: a key, a per-token price per model, and a playground for trying prompts. The public beta announcement of November 2024 offered $25 of free API credits a month during the beta. That credit was withdrawn in 2025; the forum threads that rank for "grok free tier api" are people noticing the change.

Three things still exist that get called a free tier and are not one: the playground on xAI's site (free, but it is a chat box, not an endpoint), promotional credits some platforms hand new accounts for a few days, and tutorials that route Grok through a user-pays browser SDK. None gives you a Grok API key you can call from a server for nothing.

What the Grok API costs at list, and here

xAI prices each model on its models page. The same ids are on this key at the prices below, read from the catalog when this page was built; the developer list price sits beside ours where the catalog has it.

Model	Input	Cached	Output	Developer list in / out
Grok 4.7	$0.60 (70% under)	$0.15	$1.80 (70% under)	$2.00 / $6.00
Grok 4.6	$0.60 (70% under)	$0.15	$1.80 (70% under)	$2.00 / $6.00
Grok 4.5	$0.60 (70% under)	$0.09	$1.80 (70% under)	$2.00 / $6.00
Grok 4.5 Heavy	$0.40	$0.06	$1.20	—

USD per million tokens, read at render. Every Grok id on the key is on the RelaySwitch Grok price hub with the same figures.

Grok 4.7 sits 70% under xAI's list on input here. Nothing is added on the top-up, there is no monthly minimum, and the balance does not expire.

What a $10 top-up buys

A chat turn of 2,000 tokens in and 500 out, costed at today's price:

Model	Per turn	Turns per $10
Grok 4.7	$0.0021	4,761
Grok 4.6	$0.0021	4,761
Grok 4.5	$0.0021	4,761
Grok 4.5 Heavy	$0.0014	7,142

Turns are rounded down. A longer prompt or answer costs more in proportion.

Getting a Grok API key without a monthly minimum

Open an account. An email address. No identity documents, no phone number, no developer account at xAI.
Top up. From $10, by card (Apple Pay and PayPal included) or in USDT, USDC, BTC, ETH, SOL. Nothing is added to the amount.
Create a key. Set a monthly spend limit on it if you want a ceiling; requests past it answer 402.

Base URL https://api.relayswitch.io/v1, your key (it starts with sk-relay-), the model id. Nothing else changes.

python

1from openai import OpenAI

3client = OpenAI(base_url="https://api.relayswitch.io/v1", api_key="sk-relay-...")

4reply = client.chat.completions.create(

5 model="grok-4.7",

6 messages=[{"role": "user", "content": "Say hello in five words."}],

7)

8print(reply.choices[0].message.content)

bash

1curl https://api.relayswitch.io/v1/chat/completions \

2 -H "Authorization: Bearer $RELAYSWITCH_KEY" \

3 -H "Content-Type: application/json" \

4 -d '{"model": "grok-4.7", "messages": [{"role": "user", "content": "Hello"}]}'

The Grok ids answer the OpenAI chat format, so any OpenAI SDK, Cursor, Cline or SillyTavern works with the base URL changed.

If you want Grok for roleplay or an uncensored assistant

The Grok ids here answer as xAI serves them, under xAI's usage policy. For unfiltered chat the uncensored builds on the same key are the models built for it, priced per token like everything else, and the browser roleplay client runs on the same balance.

Questions

Is the Grok API free or paid?

Paid, per token, at xAI and everywhere else that serves it. xAI's public beta of November 2024 came with monthly free credits and that allowance ended in 2025; its API page now lists per-million-token prices and a free playground, not a free API tier. Here every Grok request bills per token from a prepaid balance that starts at $10.

Is the Grok 4.5 API free?

No. Grok 4.5 is $0.60 in and $1.80 out per million tokens on this key today, read from the catalog at render. A $10 top-up is the smallest way in.

Is the Grok free tier gone?

Yes. The $25-a-month API credit of the 2024 public beta is no longer offered, and the threads that still rank for the question are answers to people asking why their credits stopped. The playground on xAI's site stays free to try prompts in; API calls are metered.

Is there a limit to the Grok free tier?

There is no free tier to limit. The limits that apply are per-token prices and, on this key, the spend limit you set on each key in the console. A monthly limit on a key stops requests with a 402 when it is reached, so a side project cannot run past what you gave it.

What is the cheapest way to call Grok 4.6 or 4.7 by API?

On this key Grok 4.7 is $0.60 in and $1.80 out per million tokens today, against xAI's list of $2.00 and $6.00, with no monthly minimum, by card or in USDT, USDC, BTC, ETH, SOL, and no identity check. The Grok price hub compares every Grok id the same way.

Originally published at https://relayswitch.io/blog/grok-free-tier-api.

Paying for an AI API in crypto

UplinkSwitch — Tue, 27 Oct 2026 00:39:30 +0000

From the Uplink Switch LLC family: this post first appeared on RelaySwitch at relayswitch.io/blog/pay-for-ai-api-with-crypto, where the figures are read live from the catalog.

October 26, 2026 · 6 min

Paying for an AI API in crypto: USDT, USDC, BTC, ETH or SOL to a prepaid balance

How a crypto top-up works here: the console quotes an amount, you send it, the balance credits in USD, and every request bills per token from it. What is stored, what is not, and what a dollar buys today.

A crypto payment here buys the same thing a card payment buys: a USD balance on your account, from which every request draws at the per-token price on the model page. There is no crypto plan, no separate price list and no markup for paying this way. This post walks through the top-up, what the account stores afterwards, and what a dollar of balance buys on the flagship models today.

Coins

USDT · USDC · BTC · ETH · SOL

Also

Card · Apple Pay · PayPal

Smallest top-up

$10

Models on the key

284

01The top-up, step by step

Open an account with an email address at the console. That is the whole sign-up; nothing else is asked for.
Press Top up, pick an amount (from $10; the preset tiers are $10, $25, $50, $100, $250, $500) and choose the coin: USDT, USDC, BTC, ETH, SOL.
The checkout shows the networks it accepts for that coin, an address and the exact amount to send. Send that amount on the network you selected.
When the payment confirms on its network, the balance credits in USD. Create a key on the keys page; it is shown once.

02What it costs to pay this way

You pay the network fee for the transfer; the amount credited is the amount received, and the checkout shows the exact total before you send. There is no percentage added for crypto and no difference in the per-token price afterwards: the balance is the same USD credit a card top-up produces.

Two things to check before sending

The network and the asset. Funds sent on a network the checkout did not list, or in another asset, may be unrecoverable, and a confirmed payment is final. The payment page has the full list of methods.

03What the account stores

For a crypto top-up: the amount, the method and the transaction hash. For a card top-up: the amount, the method and the processor's reference; card numbers never reach RelaySwitch. For every request: a usage record with the request id, the model, the token counts and the charge. The ledger of balance changes. The email address.

Not stored: request and response bodies. Prompts and completions pass through and are not written anywhere or used for training. The privacy page is the full list.

04What a dollar buys today

Prices are per million tokens, input and output separately, read from the live catalog. The last column is one chat turn of 2,000 input and 500 output tokens.

Model	In / M	Out / M	One turn	Turns per $10
Claude Opus 5.5	$2.716	$9.70	$0.0103	972
GPT-5.5	$1.80	$9.00	$0.0081	1,234
DeepSeek V4 Pro 0423	$0.792	$2.376	$0.0028	3,607
Grok 4.7	$0.60	$1.80	$0.0021	4,761
GLM 5.3	$0.3276	$1.0296	$0.0012	8,547
Kimi K3	$0.90	$4.50	$0.0041	2,469

284 models from 41 developers are on the key; the Models page lists them all.

05Then the first request

The endpoint is OpenAI-compatible. Any SDK that takes a base URL works unchanged; the model id is any row of the Models page.

bash

1curl https://api.relayswitch.io/v1/chat/completions \

2 -H "Authorization: Bearer $RELAYSWITCH_KEY" \

3 -H "Content-Type: application/json" \

4 -d '{"model": "claude-opus-5-5", "messages": [{"role": "user", "content": "Hello"}]}'

When the balance runs out, requests answer 402 until the next top-up. Nothing is charged automatically, nothing accrues, and credit does not expire.

Originally published at relayswitch.io. Prices are the live catalog's on 2026-10-27; the original page re-reads them on every load.