Your Claude or ChatGPT plan on Extent: what the terms say

Extent runs your bots on the Claude or ChatGPT plan you already pay for. Here is what Anthropic and OpenAI say about that, quoted, and where Extent stands on each point.

How Extent uses your plan

Each bot runs the official Claude Code or Codex, as Anthropic and OpenAI publish it, in a container that serves only you. You sign in yourself, on Anthropic's or OpenAI's own sign-in page. Extent seals the sign-in it gets back and uses it only for your bots.

The usage counts against your plan, under your own agreement with Anthropic or OpenAI. Extent doesn't pay for it, resell it, or pool it with anyone else's. No one else's bots run on your plan.

Anthropic

Anthropic's legal and compliance page lets a platform run Claude Code "in hosted sandboxes or other agent infrastructure", on two conditions:

The same page says the rules on sign-ins don't prevent "an end user from signing in to the unmodified Claude Code binary with their own Claude subscription, including where a platform hosts Claude Code". That is the setup Extent uses.

Two lines on that page don't fit Extent as it works today:

The page also notes that plan limits "assume ordinary, individual usage of Claude Code and the Agent SDK". A bot that works all day uses more than a person typing, and Anthropic decides what counts as ordinary.

OpenAI

For ChatGPT, you sign in with your own ChatGPT account on OpenAI's sign-in page, and your bots run Codex on your plan. OpenAI sets the rules for apps that run on a ChatGPT plan under Sign in with ChatGPT: the plan serves the person who signed in, and the app doesn't charge for model use or pool plans between people. Extent charges nothing for model use, and your plan runs only your bots.

What this means for you

We checked both providers' pages on 31 October 2026 and will update this post when they or Extent change. Questions go to hello@extent.sh.