Skip to content

fix(codex): align codex-pro context_length with client catalog max_context_window - #53

Open
jcardama wants to merge 1 commit into
router-for-me:mainfrom
jcardama:fix/codex-pro-context-length-872000
Open

jcardama wants to merge 1 commit into
router-for-me:mainfrom
jcardama:fix/codex-pro-context-length-872000

Conversation

@jcardama

@jcardama jcardama commented Sep 4, 2026

Copy link
Copy Markdown

Problem

f1d6988 refreshed codex_client_models.json, moving max_context_window for the GPT-5.6 family from 921000 to 872000 and adding gpt-6-astra at 872000. The matching context_length values in the codex-pro section of models.json were not moved, so the two files now disagree:

model models.json codex-pro codex_client_models.json max_context_window
gpt-6-astra 272000 872000
gpt-5.6-sol 921000 872000
gpt-5.6-terra 921000 872000
gpt-5.6-luna 921000 872000

Astra is the odd one out in a second way. Its 272000 is the context_window value, while its three neighbours use max_context_window semantics, so Astra currently advertises a smaller context than Sol despite being the higher-priority model.

Precedent

34a122f, "Update GPT-5.6 context length to 921000 for Codex Pro and client models", changed context_length in models.json and max_context_window in codex_client_models.json in the same commit, to the same number. This PR applies that same pairing to the value f1d6988 just landed.

Evidence

GET https://chatgpt.com/backend-api/codex/models?client_version=0.153.2 on a Pro account, captured under both originators:

slug context_window max_context_window
gpt-6-astra 272000 872000
gpt-5.6-sol 272000 872000
gpt-5.6-terra 272000 872000
gpt-5.6-luna 272000 872000
gpt-5.5 272000 272000

codex-tui and codex_cli_rs agree on every row. gpt-5.5 is included as a control: it reports 272000 for both fields and is already stored as 272000, so it needs no change and is left untouched.

Scope

Four context_length values in the codex-pro section, set to 872000. Nothing else is modified.

codex-team and codex-plus still carry 372000 for the 5.6 family and 272000 for Astra. I have left those alone deliberately: I only have a Pro credential, so I cannot observe what those plans actually return, and I would rather not change a number I cannot verify. If 372000 is a real per-plan limit then Astra's 272000 in those two sections is probably also wrong, but that needs someone on a Plus or Team account to confirm.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant