DEV Community: TrustLoud

n8n vs Zapier vs Make: What 1,000 AI Runs Would Cost (2026)

TrustLoud — Mon, 09 Nov 2026 05:58:40 +0000

Picking an automation platform for AI workflows? The three big ones meter usage so differently that the same workflow can cost ten times more on one than another. This is a cost model built from each vendor's published pricing, not a benchmark, and every source is linked inline.

TL;DR: For 1,000 runs a month of a simple AI workflow (one AI step plus two app actions), Make is cheapest at about $9 to $10.59 a month, n8n Cloud Starter costs €20 to €24 (about $23 to $27) plus a dollar or so of AI model fees, and Zapier needs its 5,000-task Professional tier at $89 to $133.50. The gap grows with AI agents: Zapier charges 3 tasks for every advanced-model AI step and every tool call, so a three-tool agent run lands on the $189 to $283.50 tier, while n8n still counts each run as one execution. Zapier is the easiest to start and has the most apps (9,000+). n8n is the cheapest per run at scale and can be self-hosted for free. Make sits in between.

Last updated: October 19, 2026. By Veer. Prices checked October 19, 2026 against Zapier’s pricing page and task usage rates, n8n’s pricing page, Make’s credits documentation and Make vs Zapier page, and OpenAI’s API pricing. Euro conversions use the ECB reference rate for October 16, 2026 (€1 = $1.1253 = £0.84385). This is a cost model built from published prices, not a measured benchmark: we haven’t run these workflows on the three platforms ourselves.

Search “n8n vs zapier vs make” and you’ll find plenty of feature checklists. What’s harder to find is the actual bill. The three tools meter usage in completely different ways, so the same workflow can cost ten times more on one than another. Below, we take two realistic AI workflows, run each one 1,000 times a month on paper, and price them on all three platforms using only the vendors’ published rates.

How do n8n, Zapier and Make charge for usage?

Zapier bills per task, Make bills per credit, and n8n bills per execution. A Zapier task is one successful action step. A Make credit is roughly one module action. An n8n execution is one full run of a workflow, however many steps it has. That single difference decides most of the cost gap.

Here’s how each vendor defines its unit:

Zapier: “Each successful action in a Zap counts as a separate task,” according to Zapier’s help center. Triggers, Filters, Paths, Formatter, Delay and Looping steps don’t use tasks. AI steps cost more: Zapier’s usage rates page lists 1 task for a standard model, 3 for an advanced model and 5 for a premium model, and each tool call an AI step makes is billed again at the same rate. A Zapier MCP tool call costs 2 tasks.
Make: “Every time a module in a scenario performs an action, it consumes credits,” says Make’s pricing FAQ. Most modules use 1 credit, and that includes the trigger that reads data in. Routers and error-handler modules are free. AI modules that run on Make’s own AI provider also charge credits for tokens. With your own OpenAI or Anthropic key, you pay Make 1 credit per operation and pay the provider for tokens.
n8n: “An execution is a single run of your entire workflow. It doesn’t matter how many steps are in the workflow or how much data it processes,” according to n8n’s pricing page. AI calls are billed by whichever model provider you connect.

What do the plans cost in October 2026?

Entry-level paid plans start at $9 a month on Make (annual billing), €20 a month on n8n Cloud (annual) and $19.99 a month on Zapier (annual, 750 tasks). The headline prices aren’t comparable, though, because each one buys a different unit. The table shows the tiers that matter for 1,000 runs.

Plan	What you get	Annual billing (per month)	Monthly billing
Zapier Free	100 tasks, two-step Zaps only	$0	$0
Zapier Professional, 750 tasks	Multi-step Zaps, webhooks	$19.99	$29.99
Zapier Professional, 2,000 tasks		$49	$73.50
Zapier Professional, 5,000 tasks		$89	$133.50
Zapier Professional, 10,000 tasks		$129	$193.50
Zapier Professional, 20,000 tasks		$189	$283.50
Make Free	1,000 credits, 15-minute minimum interval	$0	$0
Make Core, 10,000 credits	Unlimited scenarios, 1-minute scheduling	$9	$10.59
Make Pro, 10,000 credits	Adds priority execution	$16	$18.82
n8n Starter (Cloud), 2,500 executions	5 concurrent executions	€20 (about $22.51 / £16.88)	€24 (about $27.01 / £20.25)
n8n Pro (Cloud), 10,000 executions	Up to 50 concurrent executions	€50 (about $56.27 / £42.19)	€60 (about $67.52 / £50.63)
n8n Community Edition	Self-hosted, no execution limit	Free (you pay for the server)	Free

Zapier’s tier prices come from the plan data on its US pricing page, which offers billing in GBP and other currencies. Make’s annual prices are on its Make vs Zapier page. Its monthly prices come from Jet Admin’s September 26, 2026 reading of the pricing page, because Make’s credit slider doesn’t load outside a browser. n8n shows its Cloud prices in euros; the €24 and €60 monthly figures come from CloudZero’s n8n pricing breakdown and match n8n’s “Save 17%” annual discount. UK buyers should expect VAT on top of all three.

Workflow 1: What does a simple AI step cost for 1,000 runs?

About $9 to $10.59 on Make, €20 to €24 plus roughly $1 of OpenAI fees on n8n, and $89 to $133.50 on Zapier. The workflow is a common one: a new form submission or email arrives, an AI model classifies and summarizes it, a filter drops spam, then the workflow updates a CRM record and posts to Slack.

The cost math, per run and per 1,000 runs:

	Zapier	Make	n8n
Trigger	0 tasks	1 credit	Part of the execution
AI step (classify and summarize)	1 task (standard model) or 3 (advanced)	1 credit + about 1.2 credits of tokens on Make’s “Large” (GPT-5 mini) model	Part of the execution, plus provider fees
Filter	0 tasks	0 credits (Make filters sit on the link between modules)	Part of the execution
CRM update + Slack message	2 tasks	2 credits	Part of the execution
Per run	3 tasks (5 with an advanced model)	About 5.2 credits	1 execution
Per 1,000 runs	3,000 to 5,000 tasks	About 5,200 credits	1,000 executions
Plan needed	Professional, 5,000 tasks	Core, 10,000 credits	Starter, 2,500 executions
Monthly cost	$89 (annual) or $133.50 (monthly)	$9 (annual) or $10.59 (monthly)	€20 to €24 (about $22.51 to $27.01) + about $1.10 in OpenAI fees

The token assumption is 2,000 input tokens and 300 output tokens per AI call. On Make’s AI provider, the credits documentation lists GPT-5 mini at 3,616 input tokens and 452 output tokens per credit, which works out to about 1.2 credits a call. On n8n, or on Make with your own key, the same call through OpenAI’s API costs about $0.0011 at GPT-5 mini’s list price of $0.25 per million input tokens and $2.00 per million output tokens. That’s roughly $1.10 for 1,000 runs.

Zapier’s 2,000-task tier ($49 a month billed annually) is too small here, because each run burns three tasks even with the cheapest AI model.

Workflow 2: What does an AI agent with tools cost for 1,000 runs?

n8n and Make stay on their entry plans, while Zapier jumps to its 20,000-task tier at $189 to $283.50 a month. The workflow is an AI agent that reads an incoming support request, makes three tool calls (look up the customer, check an order, search a help article), then writes a reply draft and updates the ticket.

	Zapier	Make	n8n
Trigger	0 tasks	1 credit	Part of the execution
AI agent step	3 tasks (advanced model)	1 credit + tokens	Part of the execution
Three tool calls	3 × 3 = 9 tasks	About 3 credits (one module each)	Part of the execution
Draft reply + ticket update	2 tasks	2 credits	Part of the execution
Per run	14 tasks	7 credits with your own key, about 10.4 on Make’s Large model	1 execution
Per 1,000 runs	14,000 tasks	7,000 to 10,400 credits	1,000 executions
Monthly cost	$189 (annual) or $283.50 (monthly) on the 20,000-task tier	$9 to $10.59 plus about $3.10 in OpenAI fees with your own key	€20 to €24 (about $22.51 to $27.01) + about $3.10 in OpenAI fees

This run assumes 6,000 input tokens and 800 output tokens across the agent’s reasoning. On OpenAI’s GPT-5 mini that’s about $0.0031 a run. On Make’s own Large model it adds about 3.4 credits, which pushes the total just past the 10,000-credit base tier, so you’d either buy extra credits or switch to Make’s cheaper Small or Medium tier (18,080 input and 2,260 output tokens per credit), which keeps it under 8,000 credits.

Zapier gets expensive fast as the model gets smarter. With a premium model the same agent costs 22 tasks a run, which needs the 50,000-task tier at $289 a month billed annually or $433.50 monthly. Bringing your own model to Zapier cuts the agent to 6 tasks a run (1 task per AI step and per tool call), so it fits the 10,000-task tier at $129 to $193.50, plus your provider bill.

One caveat: this uses AI by Zapier inside a Zap. Zapier’s separate Agents product doesn’t use tasks at all. It has its own activity quota, according to Zapier’s help center.

Is self-hosted n8n really free?

Yes, the software is free, but the server and the upkeep are not. n8n’s hosting docs say that “without a license key, n8n runs as the free Community edition,” with no execution limit. You pay for a server, and you own updates, backups and security.

That trade-off is worth taking seriously. A self-hosted automation server holds API keys for your CRM, email and payment tools, so a missed update is a real risk. If you go this way, follow the basics in our AI agent security checklist: least-privilege API keys, human approval for risky actions, and logging. The paid self-hosted Business plan starts at €667 a month billed annually and adds SSO, Git version control and multiple environments, which most small teams won’t need.

Why is Zapier so expensive?

Zapier is expensive per run because it charges for every action step, and AI steps and tool calls cost up to 5 tasks each. You’re paying for convenience: 9,000+ app connections, a linear editor most people can use without training, and polling as often as every 2 minutes on Professional.

For a two-step Zap that runs a few hundred times a month, the difference doesn’t matter much; Zapier’s $19.99 tier covers it. The cost only bites when workflows get long, run often or lean on AI. Zapier also switches you to pay-per-task billing at a higher rate when you hit your limit, unless you turn that off, so set an alert if your volume is unpredictable.

Which is best for AI agents?

n8n is the strongest fit for AI agents if someone on your team is technical, and Make is the better pick if nobody is. n8n’s per-execution pricing means an agent that makes ten tool calls costs the same as one that makes two. Its pricing page lists an AI Agent node, an MCP client and MCP server trigger, and human approval for tool calls among its agent features.

Make’s AI Agents app works on any plan, and with your own model key each agent run costs 1 credit plus the tools it calls. Zapier’s AI steps are the simplest to set up, but the per-tool-call pricing makes busy agents the most expensive option of the three. If you’re wiring AI into your own site, our guide to connecting Claude to WordPress with MCP covers the other side of that setup, and our ChatGPT Plus vs Claude Pro vs Google AI Pro comparison helps if you’d rather use a chat app than an automation tool.

Is n8n harder to use than Zapier?

Usually, yes. n8n describes itself as “AI agents and workflows for technical teams,” and it leans on code steps, HTTP requests and JSON data. Zapier is built for people who have never written code. Make is in the middle, with a visual canvas that makes branching logic easier to follow.

n8n vs Zapier vs Make for small business: which should you pick?

Pick Zapier if you want it working today, Make if you want the lowest hosted bill, and n8n if you run lots of multi-step or AI workflows and have someone technical.

If you…	Choose	Why
Run a few simple automations and want no learning curve	Zapier	9,000+ apps, easiest editor, free plan for two-step Zaps
Want the cheapest hosted plan for moderate volume	Make	$9 a month (annual) for 10,000 credits, unlimited scenarios
Run long or AI-heavy workflows thousands of times	n8n	One execution per run, no matter how many steps
Need data to stay on your own servers	n8n (self-hosted)	Free Community edition, you control the infrastructure
Have a team that needs shared workflows and SSO	Zapier Team or n8n Business	Zapier Team starts at $69 a month (annual); n8n Business at €667

Our verdict: for most small businesses doing AI work in 2026, Make offers the best price-to-effort ratio, n8n wins once volume or complexity climbs, and Zapier is worth its premium only when ease of use matters more than cost.

FAQ

Which is cheapest for 1,000 runs a month?

Make is cheapest for simple workflows, at $9 a month billed annually or $10.59 monthly for 10,000 credits. n8n Starter costs €20 to €24 a month for 2,500 executions. Zapier needs its 5,000-task tier, at $89 to $133.50, for a three-step AI workflow.

Is self-hosted n8n really free?

The n8n Community edition is free to self-host with no execution limit. You still pay for the server and handle updates, backups and security yourself.

Why is Zapier so expensive?

Zapier charges a task for every successful action step, and AI steps and tool calls cost 1, 3 or 5 tasks depending on the model tier. Long or AI-heavy workflows multiply the task count quickly.

Which is best for AI agents?

n8n is best for AI agents if you have technical help, because each agent run counts as one execution however many tools it calls. Make is a cheaper hosted option, and Zapier is the easiest but costs the most per tool call.

Is n8n harder to use than Zapier?

Yes, for most people. n8n targets technical teams and relies more on code, HTTP requests and JSON, while Zapier is designed for non-developers.

Can n8n and Zapier work together?

Yes. n8n workflows can start from webhooks, and Zapier’s paid plans include webhooks, so a Zap can hand data to an n8n workflow and the other way around.

Sources

Zapier: pricing, task usage rates, how task usage is measured. Make: pricing FAQ, credits, Make vs Zapier, AI Provider token pricing update. n8n: pricing, hosting docs. Model prices: OpenAI API pricing. Exchange rate: ECB euro reference rates. Secondary: Jet Admin’s Make review (September 30, 2026), CloudZero’s n8n pricing guide. Related on TrustLoud: AI agent security checklist, Claude Code vs Codex, Sonnet 5.5 vs GPT-6.1 Sol cost per task.

This article was first published on TrustLoud, where it's updated as Zapier, Make and n8n change their pricing.

Claude Code vs Codex: Pricing, Usage Limits and Which to Pick

TrustLoud — Mon, 26 Oct 2026 06:08:41 +0000

Choosing between Claude Code and Codex for your terminal? This is a side-by-side of what OpenAI and Anthropic actually publish about pricing and usage limits. It's researched from their docs, not a benchmark, and every source is linked inline.

TL;DR: Both coding agents start at $20 a month: Codex comes with ChatGPT Plus, Claude Code with Claude Pro ($17 a month billed annually). The usage rules differ more than the prices. Codex on Plus is metered per five-hour window, and OpenAI publishes estimates such as 15 to 160 local GPT-6.1 Sol messages per window; ChatGPT Pro plans have no five-hour limit. Claude Code shares one pool with Claude chat, with five-hour sessions plus a weekly cap and no published message count. On the API, the default models cost the same: $2 input / $10 output per million tokens for both GPT-6.1 Sol and Claude Sonnet 5.5. Pick on workflow, then size the plan to your usage.

Last updated: October 17, 2026. Prices and limits checked against OpenAI’s ChatGPT Learn docs and Anthropic’s pricing and Claude Code docs.

Claude Code and Codex are the two coding agents most developers end up choosing between. Both run in the terminal, in the IDE, on the desktop and in the cloud. Both are bundled with a $20 chat subscription. And both have usage limits that are confusing enough to fill a lot of Reddit threads.

We haven’t run a timed head-to-head on our own codebase, so this isn’t a “we refactored 10,000 lines” review. It’s a side-by-side reading of what each company actually publishes about pricing and Claude Code vs Codex usage limits, with links, so you can choose with your eyes open.

Claude Code vs Codex at a glance

The price ladders match almost exactly; the metering doesn’t. Codex publishes per-model message estimates for Plus and removes the five-hour cap on Pro. Claude Code uses per-session multipliers on Max and a weekly cap on every paid plan.

	Codex (OpenAI)	Claude Code (Anthropic)
Entry plan	ChatGPT Plus, $20/month	Claude Pro, $20/month ($17 billed annually)
Next tiers	Pro 100, Pro 200, Pro 500 ($100 / $200 / $500)	Max 5x, Max 20x ($100 / $200)
Higher-tier multiplier	Pro 100: 5x Plus usage; Pro 200 and Pro 500: more (no published multiplier)	5x / 20x Pro usage per session
Five-hour limit	Yes on Plus; none on Pro tiers	Yes on all plans
Weekly limit	“May also apply”	Yes, on all paid plans
Shared with	ChatGPT Work and other agentic features	Claude chat, desktop, mobile, Cowork
Default coding models	GPT-6.1 Sol, GPT-6 Luna; GPT-6 Astra for hard work	Sonnet 5.5, Opus 5.5; Fable on Max
Check usage	Usage dashboard, `/status` in the CLI	Settings > Usage, `/usage` in Claude Code
Top-ups	Buy ChatGPT credits	Turn on usage credits at API rates

Sources: ChatGPT Learn pricing, About ChatGPT Pro tiers, Claude pricing, What is the Max plan?, Claude Code costs.

Which gives more usage for $20?

OpenAI is more transparent at $20, but neither gives a guaranteed number. OpenAI’s pricing doc estimates local messages per five-hour window on Plus: 5 to 45 with GPT-6 Astra, 15 to 160 with GPT-6.1 Sol and 350 to 3,000 with GPT-6 Luna. Anthropic says only that Claude Pro gives at least five times Free’s usage per session.

Both companies stress these are not hard limits. OpenAI says “tasks that look similar can consume different amounts of your allowance,” and that cloud tasks may use more than local messages. Anthropic’s pricing FAQ says usage depends on “the length and complexity of your conversations, the model you choose, and the features you use.”

The bigger structural difference is what else draws from the pool. On Claude Pro, Claude Code shares limits with every chat you have on claude.ai. On ChatGPT Plus, OpenAI says Codex shares usage with ChatGPT Work and some agentic features, but its doc doesn’t put ordinary ChatGPT chat in that pool. If you chat a lot and code a lot on one $20 plan, that matters.

Which gives more usage at $100 to $200?

ChatGPT Pro removes the five-hour wall; Claude Max multiplies it. OpenAI says “Pro plans currently have no five-hour limit,” so a long Codex session won’t stall mid-afternoon. Claude Max 5x and 20x give five or 20 times Claude Pro’s per-session allowance, but the five-hour window and a weekly cap remain. Max 5x and 20x now also include $100 and $200 a month in API credits, but Anthropic’s API credits help page says they don’t cover Claude Code, so they won’t stretch your coding allowance.

A few details worth knowing before you pay:

ChatGPT Pro 200 changed recently. New Pro 200 subscribers get 10x Plus usage, according to OpenAI’s Pro tiers article. Eligible existing subscribers keep 20x until October 29, 2026.
Claude Max 20x is the only Anthropic tier that’s cheaper per unit than Pro. It gives four times Max 5x’s session allowance for twice the price.
Claude Max includes Fable. Anthropic’s Fable article says Max can use Fable models for up to 50% of the weekly limit. On Pro, Fable runs only on paid credits.
Speed modes cost more. On ChatGPT, Fast mode uses included limits at 2.5x the Standard rate and GPT-6 Astra Ultrafast at 8x. Among individual plans, only Pro 500 gets Ultrafast.

Which is cheaper on the API?

For the default models, it’s a tie. GPT-6.1 Sol and Claude Sonnet 5.5 both list at $2 per million input tokens and $10 per million output. The differences show up in caching and at the top end, where Claude Opus 5.5 is much cheaper than GPT-6 Astra.

Model	Input / 1M	Cached input / 1M	Output / 1M
GPT-6.1 Sol	$2.00	$0.10	$10.00
Claude Sonnet 5.5	$2.00	$0.10 (cache read)	$10.00
Claude Opus 5.5	$4.00	$0.20 (cache read)	$20.00
GPT-6 Astra	$10.00	$1.00	$50.00
Claude Fable 5.1	$10.00	$0.25 (cache read)	$50.00
GPT-6 Luna	$0.10	$0.01	$0.50

Sources: Claude pricing; OpenAI API prices as listed in our GPT-5.5 retirement guide, checked against OpenAI’s API pricing page. Anthropic halved Sonnet 5.5 cache reads from $0.20 to $0.10 per million tokens on October 7, 2026, matching GPT-6.1 Sol’s cached input price; our Sonnet 5.5 vs GPT-6.1 Sol comparison goes deeper on API costs. Anthropic also charges for cache writes ($2.50 per million on Sonnet 5.5); OpenAI’s Codex credit billing has no separate cache-write charge.

Running either agent on an API key has trade-offs. OpenAI says Codex with an API key loses cloud features such as GitHub code review and Slack. Anthropic says API-key use in Claude Code is “pay-as-you-go, billed per token,” with “no hard stop,” so set a spend limit in the Console.

Codex cloud tasks vs Claude Code local: what’s the difference?

Both now do both. Codex runs locally (CLI, IDE extension, desktop app) and in Codex Cloud, which runs tasks in OpenAI-hosted environments and can review pull requests when you tag @Codex on GitHub. Claude Code also runs in the cloud: Anthropic’s cloud docs say cloud sessions are available on Pro, Max and Team plans.

The usage rules differ slightly. OpenAI warns that cloud tasks “may use more of your allowance than local messages,” and GitHub code reviews have their own usage category. In Claude Code, cloud and local sessions draw from the same plan limits as everything else.

Is Claude Code or Codex better for big refactors?

There’s no independent, current answer we’d stand behind. Model rankings shift every few weeks, and benchmark results depend heavily on the harness. What’s documented is how each tool handles large changes: both support plan-first workflows, subagents and project instruction files, so either can do the job with good prompting.

Anthropic’s own Claude Code guidance suggests planning with Opus and executing with Sonnet, built in as /model opusplan. OpenAI’s models page points complex coding to GPT-6.1 Sol and the hardest work to GPT-6 Astra. In practice, run the same real refactor through both on a branch and compare the diffs.

Can you use Claude Code and Codex together?

Yes, and plenty of developers do. The easiest setup is one shared instruction file: Codex reads AGENTS.md, and Anthropic’s memory docs say Claude Code can read a repository’s AGENTS.md on its own or alongside CLAUDE.md.

A common split:

Write and iterate locally in whichever agent you prefer.
Review with the other one. Tag @Codex on a GitHub pull request, or run Claude Code’s review tools, so a second model checks the first one’s work.
Keep both on $20 plans until one of them hits its limit regularly, then upgrade only that one.

Two $20 plans also give you two separate usage pools, which can be cheaper than jumping one of them to $100.

Verdict: which should you pick?

Pick Codex (ChatGPT Plus or Pro) if you want the clearest published limits, use ChatGPT heavily for other things, or want long uninterrupted sessions: Pro tiers have no five-hour cap.
Pick Claude Code (Claude Pro or Max) if you prefer Claude’s models and terminal workflow, want Opus at a lower API price than Astra, or want Fable included on Max.
Try both for a month at $20 each if you’re unsure. It’s the cheapest way to find out which one fits your codebase.

FAQ

Which gives more usage for $20, Claude Code or Codex?

Neither guarantees a number. OpenAI estimates 15 to 160 local GPT-6.1 Sol messages per five-hour window on ChatGPT Plus, while Anthropic says Claude Pro gives at least 5x Free’s usage per session, shared with chat.

Does Codex have a five-hour limit on ChatGPT Pro?

No. OpenAI says Pro plans currently have no five-hour limit, though weekly limits may apply. Plus and Standard Business seats do have five-hour windows.

Is Claude Code or Codex cheaper on the API?

For default models they’re the same: GPT-6.1 Sol and Claude Sonnet 5.5 both cost $2 input and $10 output per million tokens. Claude Opus 5.5 ($4/$20) is cheaper than GPT-6 Astra ($10/$50).

Can I use Claude Code and Codex together?

Yes. Both can read an AGENTS.md file, so one set of project instructions works for both, and many developers use one agent to write and the other to review.

How do I check my Claude Code and Codex limits?

Run /usage in Claude Code or check Settings > Usage on claude.ai. For Codex, use the usage dashboard in ChatGPT’s Codex settings or /status in the Codex CLI.

Sources checked October 11, 2026. Both companies change limits often; we’ll update this comparison when they do.

This article was first published on TrustLoud, where it's updated as OpenAI and Anthropic change their plans and limits.

How to Use OpenAI Dots: Setup Guide and Permission Lockdown

TrustLoud — Mon, 19 Oct 2026 05:58:58 +0000

If you're building with (or around) OpenAI's new always-on agents, this is the setup and permissions walkthrough for dots. It's researched from OpenAI's own documentation, and the sources are linked throughout.

TL;DR: To use OpenAI dots you need ChatGPT Pro ($100, $200 or $500 a month, outside the UK, EEA and Switzerland), a ChatGPT Business Premium seat, or an Enterprise workspace where an admin has switched dots on. Plus and Go don't qualify. As of October 10, 2026, you can create your dot in the ChatGPT mobile app as well as on desktop. Before you give it real work, connect only the apps it needs, set plugin permissions to read-only where you can, add custom rules for sending and deleting, and learn the difference between Pause, Activity and Scheduled, because pausing a dot doesn't cancel its recurring tasks.

Last updated: October 10, 2026.

OpenAI launched dots on September 29 as always-on agents inside ChatGPT. This guide covers how to use OpenAI dots safely once one shows up in your account: setup, what to connect, and which switches to flip before you hand it your inbox.

Everything here comes from OpenAI's own documentation on learn.chatgpt.com, checked on October 10, 2026, plus reporting where noted. We haven't run a dot ourselves, so this is a researched walkthrough of the documented setup, not a review of how well dots perform.

What a dot actually is

A dot is a personal agent that keeps working between your conversations. According to OpenAI's Meet dots page, it runs on GPT-6 Astra, has its own computer and browser in the cloud, and can keep working while your own devices are off.

You give it ongoing responsibilities rather than one-off prompts, such as keeping an offsite plan current. It decides when to follow up, comes back to you for decisions, and can hand jobs to Codex or ChatGPT Work.

Use this	When you want
Chat	An answer, an explanation or a quick draft
ChatGPT Work	A one-off or scheduled task, such as research or a file
Codex	To build, debug or review software
A dot	Something to own an ongoing job and follow up as things change

Source: OpenAI's comparison table on the Meet dots page.

Who can use OpenAI dots right now

Dots are on three plan types only, and they're rolling out gradually, so an eligible account may not see them on day one. OpenAI's access section lists:

Plan	Price (USD)	Dots?
Free, Go ($8), Plus ($20)	$0 to $20 a month	No
Pro 100, Pro 200, Pro 500	$100, $200 or $500 a month	Yes, 18+, outside the EEA, UK and Switzerland
Business, Premium seat	$125 a month, or $100 billed annually	Yes, rolling out worldwide
Business, Standard seat	$25 a month, or $20 billed annually	No
Enterprise, Edu, Healthcare	Contact sales	Beta, off until an admin enables it

Prices from OpenAI's ChatGPT pricing page and Business pricing page, checked October 10, 2026. Business needs at least two seats.

Your first dot is included at no extra cost. OpenAI's launch post says extra dots and more speed or workload will come "in the future", with no prices yet.

In Britain or Europe, no Pro tier gets you a dot; our OpenAI dots UK and EU guide covers the Business Premium route. Choosing a Pro tier? See our ChatGPT Pro $100 vs $200 vs $500 comparison.

What counts toward your limits

Talking to your dot doesn't use your ChatGPT usage limits. Tasks it starts in ChatGPT Work or Codex do count toward those products' limits as usual. OpenAI says plans include an allowance for "deeper work, with extended limits for the first month after launch", so expect the terms to tighten once that first month ends.

How to use OpenAI dots: set up your dot step by step

Setup takes three steps, and you can now do it from your phone. OpenAI's October 5–9 update added mobile setup, so older guides that say "desktop only" are out of date.

Open dots. Use the latest ChatGPT mobile app, the desktop app, or ChatGPT in a desktop browser. Mobile web isn't supported.
Name it and pick a look. Choose a name, shape, color, eyes, glasses and accessories. Your dot starts with a handle like @yourname-dot, and naming it changes the handle (a dot named Alfred becomes @yourname-alfred).
Connect plugins, or skip. Add the apps you want it to use, such as email, calendar and files. You can skip this and add them later, which is what we'd suggest (more on that below).

After setup, the same dot follows you across phone, desktop and web.

Optional: connect your computer

You don't need to. The dot has its own cloud computer. If you want it to use local files, your signed-in browser or local Codex tasks, the computers and apps guide says to open the dot's profile in the ChatGPT desktop app, go to Computers, find Your computer and select Allow access, then confirm.

Only one personal computer can be connected at a time, and it has to be online with the ChatGPT app open while the dot uses it. Revoke access removes the connection. An "Offline" status doesn't.

Optional: add Slack or Teams

Open the dot's profile and select Add. In Slack you can DM your dot or mention it in a thread, and by default it only answers you. Adding it to a channel doesn't make it watch that channel; you have to ask. Texting is "coming soon", per the messaging page.

How to lock down a dot's permissions

Each connection is separate: connecting Slack doesn't connect your inbox, and Gmail doesn't connect your computer. So you can give a dot exactly what it needs and nothing more.

OpenAI also runs an automatic review before any action that could affect your accounts or share information, deciding whether the dot can proceed, needs approval, or must hand the step to you (changing a password always goes to you). On top of that, you have three layers of control.

Layer 1: plugin permissions

Your existing ChatGPT app permissions apply to your dot. OpenAI's example is allowing a dot to read email without allowing it to send email. Start every plugin at the narrowest setting that still lets the dot do its job. You can review this from Custom rules via Open Plugins, under Plugin permissions.

Layer 2: custom rules

Go to Settings > Personalization, then Custom rules under Permissions, and select Add. Describe an action and pick one of four behaviors, as listed in OpenAI's controls guide:

Rule	What it does	Good for
Take action without asking	Acts with no approval step	Low-risk, reversible jobs, such as updating your own checklist
Take action when you say so	Acts only when you explicitly ask; otherwise asks first	Posting a draft you've already reviewed
Ask before taking action	Always asks for approval	Emailing clients, booking anything, sharing files
Hand off to you	Makes you do it yourself	Deleting shared files, payments, account changes

A sensible starting set for most people:

Ask before taking action for sending any email or message to someone other than you.
Ask before taking action for sharing files or changing who can see them.
Hand off to you for deleting files, emails or calendar events.
Hand off to you for anything involving money.

Two caveats from OpenAI's docs: custom rules are "instructions your dot tries to follow, and it can make mistakes," and they can't grant access that a plugin permission blocks. In a work workspace, an admin can switch custom rules off entirely.

Layer 3: plain instructions

For style and process, just tell it. OpenAI notes that asking your dot to draft replies "doesn't give it permission to send them." A line like "Show me every draft before it goes out, and never reply in a channel without checking with me" is worth adding on day one.

Website sign-ins

When a site needs a login, the dot sends a private sign-in form, so your credentials go to its cloud browser outside the chat. Save to Passwords is optional, and reusing a saved login always needs your confirmation.

How to pause, stop or delete a dot

This is where people are most likely to get caught out, because the three controls do different things:

Pause stops the dot's current main task only. It doesn't stop delegated tasks or cancel scheduled runs. Resume restarts it.
Activity (in the desktop app, from the dot's profile) is where you open and stop individual delegated tasks.
Scheduled is where you disable or delete recurring tasks.

Ending a voice call doesn't necessarily stop work, and nothing undoes actions already taken. Delete removes the dot permanently, but it doesn't reverse changes made in connected apps or recall messages already sent.

When you set a recurring task, OpenAI recommends giving a time zone and an end date, then asking the dot to confirm the schedule. For example: "Check the planning channel each weekday at 9 AM Eastern for the next four weeks. Message me only if a deadline is at risk. Confirm the schedule."

Why your dot might not be working

If you can't see dots, or your dot keeps failing, these are the documented reasons:

Not rolled out yet. OpenAI says you may not see dots immediately even on an eligible plan.
Wrong plan or region. Plus and Go don't qualify, and Pro is excluded in the UK, EEA and Switzerland.
Mobile web. It isn't supported. Use the app.
Enterprise. An admin must enable Use dots (Beta) under Workspace settings > Permissions & roles, per the admin guide.
A website blocks the cloud browser. Some do. You can ask the dot to try your connected computer instead.
An expired connection. Open the dot's request and reconnect the plugin.

There are wider teething problems too. The Register reported on October 7 that users were seeing task failures, connection issues and dots that disappeared. OpenAI's October 9 update fixed several bugs, including Outlook setup getting stuck.

Verdict: how to start with a dot

Treat your first week as a trial with a short leash. Connect one or two apps with read-only access, give the dot a single ongoing job that you'd normally check yourself (a project tracker or a weekly inbox summary), and require approval for anything that leaves your account. Review what it does in Activity every day or two. Widen its permissions only once you've seen that its drafts and decisions hold up.

On Plus and tempted to upgrade just for this? We'd wait: the cheapest route is $100 a month and the extended launch limits end after the first month. For more OpenAI changes, see our AI news this week roundup.

FAQ

Can ChatGPT Plus users get dots?

No. OpenAI lists only Pro ($100, $200 or $500 a month), Business Premium seats and Enterprise, Edu and Healthcare workspaces as eligible. OpenAI says it plans to expand dots to more users, but it hasn't said when or to which plans.

How much do OpenAI dots cost?

Your first dot is included at no extra cost with an eligible plan. The cheapest personal route is ChatGPT Pro at $100 a month. A Business Premium seat is $125 a month, or $100 billed annually, and Business needs at least two seats. Extra dots aren't priced yet.

Can I set up a dot on my phone?

Yes, since OpenAI's October 5–9 update. You can create, name and customize a dot and connect plugins in the latest ChatGPT mobile app. Mobile web isn't supported, and connecting your own computer still has to be done from the desktop app.

How do I stop my dot from sending emails?

Set the email plugin to read-only in its plugin permissions, then add a custom rule under Settings > Personalization > Custom rules set to "Ask before taking action" or "Hand off to you" for sending messages. Telling it to draft only also helps, but drafting doesn't grant permission to send.

Does pausing a dot stop everything?

No. Pause stops only the current main task. You have to stop delegated tasks in Activity and disable recurring tasks in Scheduled separately, and stopping work doesn't undo actions already completed.

Why can't I see dots in ChatGPT?

The most common reasons are that the rollout hasn't reached your account yet, your plan isn't eligible, you're on Pro in the UK, EEA or Switzerland, or you're using mobile web. In an Enterprise workspace, an admin has to enable dots first.

Sources checked on October 10, 2026. We'll update this guide when OpenAI changes dots pricing, plans or controls.

This article was first published on TrustLoud, where it's kept up to date as OpenAI changes dots.