### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? ▎ ▎ ▎ Claude Max (claude.ai OAuth). Same resumed session, same 663,871-token prompt (~1.2 MB of @-injected files + history), same account, same day.

[BUG] Title: 2.1.26x rejects >200K prompts on Max subscription ("Prompt is too long"); 2.1.257 serves the same 664K prompt

Status Open

Reported on v2.1.257

Maintainer reply None cached

Activity 0 comments · opened Sep 12, 2026

Preflight Checklist

[x] I have searched existing issues and this hasn't been reported yet
[x] This is a single bug report (please file separate reports for different bugs)
[x] I am using the latest version of Claude Code

What's Wrong?

▎
▎
▎ Claude Max (claude.ai OAuth). Same resumed session, same 663,871-token prompt (~1.2 MB of @-injected files + history), same account, same day.
▎ - 2.1.257: served by claude-fable-5-1, 663,871 input tokens.
▎ - 2.1.269: API invalid_request: Prompt is too long on claude-fable-5-1, claude-fable-5, and claude-opus-4-8[1m] — the last of which carried this same conversation at
▎ 679,998 tokens on 2.1.197 (2026-07-01).
▎ - Not set: CLAUDE_CODE_DISABLE_1M_CONTEXT, ANTHROPIC_BETAS. CLAUDE_CODE_DISABLE_UNKNOWN_MODEL_WINDOW_ENFORCEMENT=1 made no difference. autoCompactEnabled: false.
▎ - Downgrading to 2.1.257 fixes it immediately. Suspect the 1M-context/Fable changes in v2.1.260–2.1.261. Same class as #43989 / #34158 / #11281.

What Should Happen?

Letting me continue a >200k token conversation the previous version did fine

Error Messages/Logs

⎿  Context limit reached · /compact or /clear to continue · auto-compact is off ·                                                                                        
       /config to turn it on

Steps to Reproduce

Take the same >200k conversation without compacting and try to continue, with both installed, one launched with
~/.local/bin/claude-2.1.257

or equivalent, instead of claude (where it will work)

and another with ~/.local/bin/claude-2.1.269

where it will request a compact, essentially negating the 1M context feature of the models

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

2.1.257

Claude Code Version

2.1.269

Platform

Anthropic API

Operating System

Ubuntu/Debian Linux

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_

View original on GitHub ↗