# Why We Don't Support Claude Code (And Won't)

Anthropic prohibits third-party services from routing requests through Claude subscription credentials. The policy, the January 2026 crackdown, and what it means.

*Published 2026-06-12 · https://proxyllm.ai/blog/why-no-claude-code-support*

We get this question weekly: "Codex Hosted works with a ChatGPT subscription, can you do the same with my Claude Max plan?" No. Anthropic prohibits exactly that arrangement, in writing, and enforces it. This article lays out the policy, the enforcement history, and the principle that decides what we build.

## What Anthropic's policy says

Anthropic's Claude Code documentation is unambiguous. OAuth authentication "is intended exclusively for purchasers of Claude Free, Pro, Max, Team, and Enterprise subscription plans and is designed to support ordinary use of Claude Code and other native Anthropic applications." And then the sentence that settles it:

> "Anthropic does not permit third-party developers to offer Claude.ai login or to route requests through Free, Pro, or Max plan credentials on behalf of their users." (code.claude.com/docs/en/legal-and-compliance)

A hosted Claude Code service signed in with your subscription would be precisely that: a third-party developer routing requests through plan credentials on behalf of users. Anthropic adds that it "reserves the right to take measures to enforce these restrictions and may do so without prior notice."

## The January 2026 crackdown

This is not theoretical policy. On January 9, 2026, Anthropic deployed server-side checks that cut off third-party tools, including OpenCode, Cline, RooCode, and Goose, from authenticating with Claude subscription OAuth tokens. No advance warning. Working setups died overnight, and the affected users' subscriptions kept billing while their tooling broke.

In February 2026, Anthropic clarified the position: "Using OAuth tokens obtained through Claude Free, Pro, or Max accounts in any other product, tool, or service, including the Agent SDK, is not permitted and constitutes a violation of the Consumer Terms of Service."

In June 2026, Anthropic introduced a metered Agent SDK credit for subscribers, which restores limited personal programmatic use (`claude -p`, the Agent SDK in your own projects) against a monthly credit. Note what did not change: the prohibition on third-party services offering Claude login or routing through plan credentials for their users. The hosted-service shape remains banned.

## Why the OpenAI side is different

The contrast is vendor policy, not technical capability:

|                                          | OpenAI Codex                                           | Claude Code                                                    |
| ---------------------------------------- | ------------------------------------------------------ | -------------------------------------------------------------- |
| Non-interactive mode                     | `codex exec`, documented for scripts and CI            | `claude -p`, allowed for your own use against a metered credit |
| Plan-backed usage                        | Included in ChatGPT plans, ChatGPT sign-in recommended | Subscription OAuth restricted to Anthropic's own apps          |
| Headless sign-in                         | Device-code flow documented for servers                | Not offered for third-party use                                |
| Third-party services on plan credentials | Not addressed; account rules apply                     | Expressly prohibited                                           |

OpenAI documents the building blocks Codex Hosted uses and includes Codex in the plans; the remaining uncertainty is discretionary, which is why [our terms](/terms) say OpenAI has the final call. Anthropic answered the question in advance: no. Details on the OpenAI side are in [is Codex Hosted against OpenAI's terms?](/blog/is-codex-hosted-against-openai-terms)

## The principle

We build on documented functionality and we do not build against explicit vendor policy. A Claude Code lane would put your Anthropic account at risk of the January 2026 treatment, and a feature that can vanish overnight is not a feature, it is a liability with a UI. Vendors set the rules for their own services; our job is to be straight with you about what those rules are.

## If you want Claude models anyway

You can run Claude models through ProxyLLM today, the supported way: bring your own OpenRouter key and request `anthropic/` models through the same OpenAI-compatible endpoint, no markup, full request logs. See the [Claude integration](/integrations/claude). Your OpenAI-bound workloads are where the subscription economics live, through [Codex Hosted](/); the [calculator](/calculator) shows what that side saves.