openai / openai/codex

GPT-5.6 Sol still receives 272K max_context_window after long-context rollout; Terra and Luna receive 872K

Open
#39,144 13 comments 7 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug context
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

What version of Codex are you using?

Codex Desktop and Codex CLI on Windows. The issue is observed after the recent long-context rollout.

Codex Desktop: 26.814.41407
Codex CLI: 0.147.0

Which models are affected?
  • gpt-5.6-sol (affected)
  • gpt-5.6-terra (comparison)
  • gpt-5.6-luna (comparison)
What issue are you seeing?

After the recent long-context rollout, the server-delivered Codex model catalog still gives GPT-5.6 Sol a smaller context profile than Terra and Luna on the same ChatGPT account.

A fresh /models fetch returned the same ETag for all three models:

gpt-5.6-sol:   max_context_window = 272000
gpt-5.6-terra: max_context_window = 872000
gpt-5.6-luna:  max_context_window = 872000

Sol consequently returns to the usual ~258,400 effective context window, while Terra and Luna receive the expanded profile.

This does not appear to be stale local cache or an ignored configuration:

  1. All values came from the same fresh model-catalog response and ETag.
  2. Temporarily raising Sol's max_context_window in models_cache.json made Codex honor a larger model_context_window across turns.
  3. Once the cache TTL elapsed and /models refreshed, Sol reverted to 272000 and the effective thread context returned to ~258.4K.

This points to server-delivered catalog metadata or an account-level rollout/entitlement mismatch specific to gpt-5.6-sol.

What is the expected behavior?

If long-context access is enabled for the account and the same catalog response grants Terra and Luna 872000, GPT-5.6 Sol should receive the applicable expanded context profile as well.

If Sol intentionally has a different Codex-specific limit, Codex should document that distinction and the rollout criteria.

Additional information
  • Feedback ID: 01a01140-1caa-7ad0-8811-573c72ac3ce7
  • Related historical reports: #31860 and #32806. This report concerns the post-rollout, model-specific discrepancy rather than the earlier general 272K cap.
  • Sanitized catalog excerpts, ETags, and timestamps can be provided if needed.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by examining the fresh /models response and the models_cache.json behavior for gpt-5.6-sol, gpt-5.6-terra, and gpt-5.6-luna. Trace where max_context_window is obtained and applied in Codex CLI or Desktop, then verify whether the catalog discrepancy is account rollout metadata or an intentional model limit. Done when Sol receives the applicable profile or the distinction is documented.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
api, backend
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.