What is Nous Portal?
Nous Portal launched on April 27, 2026 as Nous Research’s subscription product for users who want to run Hermes Agent without managing API keys for every model and tool provider. One subscription replaces logins to a model gateway plus Firecrawl, FAL, Krea, Browser Use, OpenAI audio, and Modal — Portal routes all of those behind a single authenticated session. Since then it has also grown a hosting product, Hermes Cloud, covered below.
The free tier is $0/mo and carries $0 in monthly credits — it gives you the free-model catalog and standard rate limits only. Every paid tier adds a credit budget with a 10% bonus, the bundled Tool Gateway, and higher rate limits.
Portal plans
- Free: $0/mo, $0 monthly credits (free models only), no rollover, 50 RPM / 500K TPM
- Plus: $20/mo, $22 in credits, $10 rollover cap, 400 RPM / 4M TPM
- Super: $100/mo, $110 in credits, $50 rollover cap, 800 RPM / 8M TPM
- Ultra: $200/mo, $220 in credits, $100 rollover cap, 1,600 RPM / 16M TPM
You can also skip the subscription and top up credits directly (custom amount via Stripe); pay-as-you-go keys without a plan run at 180 RPM / 720K TPM. Plan details live on the Portal info page. Prices verified August 1, 2026.
Why Nous Portal Exists
Running a capable AI agent on your own infrastructure used to mean signing up for and paying multiple providers: a model provider (Anthropic, OpenAI), a web search provider (Tavily or Exa), an image generation provider (FAL), a TTS provider, and optionally a browser automation provider. Each one has its own dashboard, billing, and quota. For users who just want to run Hermes Agent and use it, that’s a lot of plumbing.
Nous Portal collapses that into one subscription. You pay one bill, you authenticate once, and Hermes Agent uses your Portal subscription to call models and tools automatically.
Self-host it, put it on Portal’s own Hermes Cloud, or deploy a managed Hermes Agent here in about 30 seconds — no VPS, no Docker, no credit minimum.
What’s Included
Models
Portal lists 248 models plus 29 embedding models — Anthropic Claude, OpenAI GPT, Google Gemini, Meta Llama, Mistral, DeepSeek, Qwen, xAI Grok, and Nous Research’s own Hermes series, among others. Nous states plainly on its own info page that this catalog is powered by OpenRouter, and the model links on the Portal site point at openrouter.ai. Worth knowing before you compare: for third-party models you are paying Portal to sit in front of OpenRouter, not to replace it.
Nous’s own models are the exception — Hermes-4.3–36B, Hermes-4–70B and Hermes-4–405B (128K context each) run on Nous’s own inference backend at inference-api.nousresearch.com/v1, which is OpenAI-compatible. That endpoint also supports x402 payments in Solana USDC (beta), which lets you make pay-per-request calls with no account and no API key at all.
Some models are free for Portal subscribers; others bill against your monthly credit budget.
Tool Gateway (paid tiers)
Paid Portal subscribers get automatic access to bundled tools, with no extra API keys needed:
- Web search & scraping — Firecrawl, billed at $0.0005 per credit
- Image generation — FAL (FLUX 2 Pro, Nano Banana Pro, Ideogram v3, Recraft v4, Qwen Image) and Krea, roughly $0.005-$0.26 per image
- Video generation — FAL, covering Veo 3.1, Kling v3, Seedance 2.0, PixVerse v6 and LTX-2 ($0.0017-$0.44 per unit)
- Speech in and out — OpenAI audio: Whisper transcription at $0.0063 per audio minute, plus GPT-4o transcribe and TTS billed per token
- Browser automation — Browser Use at $0.0011 per session minute, plus $4.20/GB proxy bandwidth
- Code execution — Modal sandboxes at $0.0495 per CPU-hour and $0.0084 per GiB-hour
These tools route through Nous Research’s gateway. Hermes Agent picks them up automatically when you authenticate with a Portal subscription, no per-tool config required. Every tool call bills against the same credit balance as your models, so a heavy image or video session eats the same budget your inference does.
xAI Grok integration (May 2026)
On May 15, 2026, Nous Research and xAI shipped a Grok integration that lets X Premium+ subscribers use their existing Grok subscription inside Hermes Agent — without paying separately for Grok API access. Grok 4.3, Grok TTS, and Grok Imagine all become available through the OAuth flow. See the launch coverage for details, or the upstream xAI Grok OAuth guide.
Hermes Cloud — Nous now hosts the agent too
The biggest change since Portal launched: it is no longer only a key-and-credits product. Hermes Cloud deploys a hosted Hermes Agent on a dedicated instance with its own workspace and dashboard, billed hourly from your Nous credit. It is currently in preview.
You pick a size, and the instance price covers compute and storage only — inference and tool usage bill separately on top:
- Small: 1GB RAM, 2 vCPU, 5 concurrent sessions. $0.29/day running (about $8.70/mo), $0.03/day stopped
- Medium: 2GB RAM, 4 vCPU, 10 concurrent sessions. $0.56/day running (about $16.80/mo), $0.03/day stopped
- Large: 4GB RAM, 8 vCPU, 20 concurrent sessions. $1.09/day running (about $32.70/mo), $0.03/day stopped
Stopping an instance drops it to the storage-only rate and keeps your data, so an idle agent costs about $0.90/mo to keep alive. Deploying requires an active subscription or a credit balance — the Cloud page states a $10 minimum, while the info page’s getting-started steps say $2, so budget for the higher number. Nous picks the region closest to you from Amsterdam, Stockholm, Mumbai, Paris, Dallas, Frankfurt, Los Angeles, London, Tokyo, Chicago, Singapore, San Jose, Sydney and Toronto.
Feature-wise it is what you would expect from a first-party host: one-click deploy, natural-language scheduling, Telegram / Discord / Slack / email / CLI channels against one shared memory, and an isolated hardened container per agent. On billing, the one rule Nous states unambiguously is the two-rate split: you pay the running rate while the instance is on and the storage rate while it is stopped.
Watch: Nous Portal in 90 seconds
One note before you watch: this was recorded in July 2026, when Portal advertised a 300+ model catalog and five Tool Gateway backends. Nous has since restated the catalog as 248 models and added a sixth backend (Modal sandboxes) plus Hermes Cloud. The walkthrough of the setup flow is still accurate; the numbers above it are the current ones.
https://www.youtube.com/watch?v=VvkBFyF14kk
How to Connect Nous Portal to Hermes Agent
Two paths, depending on whether you’re self-hosting Hermes Agent or running it on managed infrastructure.
Self-hosted Hermes
- Subscribe at portal.nousresearch.com
- Generate a Portal API key from the dashboard
- In Hermes Agent, run hermes with no arguments to enter interactive setup, or edit your config.yaml
- Select Nous Portal as the provider and paste the API key
- Hermes will list available models from your subscription — pick your default
Once configured, the Tool Gateway features (search, image gen, TTS, browser) are picked up automatically — you don’t need separate API keys for Firecrawl, FAL, or Browser Use anymore.
Do You Actually Need Portal?
Depends on where Hermes runs.
You self-host Hermes
Portal is worth it. One login replaces a model provider, Firecrawl, FAL, a TTS key and Browser Use — and the Tool Gateway comes with it. The setup steps are above.
You’d rather not run a server
You now have two managed options, and it is worth being straight about that. Hermes Cloud is Nous’s own, from $0.29/day plus inference — the tightest possible integration with Portal, though it is in preview and you need a credit balance or a subscription before you can deploy anything. OpenClaw Launch’s Hermes hosting is the alternative: deploy in about 30 seconds with no credit minimum and no card for the free trial, run either Hermes Agent or OpenClaw from the same account, and reach channels Hermes Cloud doesn’t list — WhatsApp and WeChat included. Bring a Portal key if you want, or use our managed models and skip Portal entirely.
Nous Portal vs OpenRouter vs Per-Provider Keys
Nous Portal
Pros: One auth; bundled tools; free models for subscribers; tight Hermes integration
Cons: Locked to Nous as the gateway; routing markup on paid model calls
OpenRouter
Pros: One API key for 300+ models; granular per-model spend control; it is what Portal routes third-party models through anyway
Cons: No bundled tools (search / image / video / speech / browser need separate keys)
Per-provider keys
Pros: Direct billing relationship; no gateway markup; full provider features
Cons: 5+ separate sign-ups, dashboards, and bills to manage
The right answer depends on shape of usage. For a single user running Hermes full-time, Nous Portal’s bundled tools usually win on convenience. For teams running production agents at scale, direct provider keys often come out cheaper.
Frequently Asked Questions
Is Nous Portal free?
There is a free tier, but be clear on what it gives you: $0/mo with $0 in monthly credits, capped at 50 RPM / 500K TPM, and limited to the free-model catalog. It is not a trial of the paid product — the Tool Gateway and the paid model catalog both sit behind a subscription or a credit top-up. Plus at $20/mo is the real entry point, and it returns $22 in credits.
Do I need a Nous Portal subscription to use Hermes Agent?
No. Hermes Agent is open source and free; you can run it with any provider you like — OpenRouter, direct Anthropic / OpenAI / Google keys, local models via Ollama, or Nous Portal. Portal is one option among many, not a requirement.
What models are free on Nous Portal?
Nous Research rotates a list of free-for-subscribers models, typically including their own Hermes model series and select open-weight models from other labs. The current list is in the Portal dashboard — it changes from month to month.
Can I use my X Premium+ Grok subscription with Hermes through Portal?
Yes, since May 15, 2026. The xAI Grok OAuth flow connects your existing X Premium+ subscription to Hermes Agent, giving you Grok 4.3, Grok TTS, and Grok Imagine without separate API billing. See the upstream OAuth guide.
What is Hermes Cloud, and how is it different from a Portal subscription?
A Portal subscription buys models, tools and credits. Hermes Cloud is the hosting layer on top: it runs the agent itself on a Nous-managed instance, billed hourly from the same credit balance, from $0.29/day for the Small size plus whatever inference and tools you use. You can hold a Portal subscription without ever using Hermes Cloud — plenty of people just point a self-hosted Hermes at their Portal key.
Is Hermes Cloud cheaper than managed Hermes hosting elsewhere?
On the instance line alone it looks inexpensive, but compare the whole bill. Hermes Cloud’s day rate excludes inference and tool usage, and you need a credit balance or an active subscription before you can deploy at all. It is also still in preview. If you want to try managed Hermes without funding a balance first, our Hermes hosting has a free trial and no credit minimum — and you can compare the two honestly from there.
Does OpenClaw Launch support Nous Portal?
OpenClaw Launch’s Hermes hosting runs Nous Hermes models on managed OpenClaw infrastructure today via OpenRouter — no Portal key needed. Native Hermes Agent framework hosting is also live today — pick Hermes as your framework in the configurator and deploy in about 30 seconds.
Originally published at openclawlaunch.com/guides/nous-portal, where it is kept up to date.
Top comments (0)