Skip to content

1m context window gone after Gpt 5.5 Release. #19208

Closed as not planned
Closed as not planned

Description

@Zakisb

What version of the Codex App are you using (From “About Codex” dialog)?

26.422.20832

What subscription do you have?

Pro plan

What platform is your computer?

Mac os

What issue are you seeing?

I've enable 1m context window via codex config since the release of GPT 5.4 and it was working exceptionally well. today Codex announces GPT 5.5 stating it supports 1m Context window but weirdly thought it's gone in the codex app.

Image

What steps can reproduce the bug?

Just launch the codex app and enable 1m context window in config.toml file

What is the expected behavior?

Having 1m context window automatically enabled

Additional information

here's my current config.toml file

model = "gpt-5.5"
model_reasoning_effort = "xhigh"
model_context_window = 1000000
model_auto_compact_token_limit = 900000
personality = "pragmatic"

Activity

  1. added
    appIssues related to the Codex desktop app
    on Apr 23, 2026
  2. added
    bugSomething isn't working
    contextIssues related to context management (including compaction)
    configIssues involving config.toml, config keys, config merging, or config updates
    on Apr 23, 2026
  3. BillDuke13 commented on Apr 23, 2026

    @BillDuke13

    cli v0.124.0 has this issue too

  4. cccat6 commented on Apr 23, 2026

    @cccat6

    I read the release page https://openai.com/index/introducing-gpt-5-5/ it says

    In Codex, GPT‑5.5 is available for Plus, Pro, Business, Enterprise, Edu, and Go plans with a 400K context window. GPT‑5.5 is also available in Fast mode, generating tokens 1.5x faster for 2.5x the cost.
    

    Which 400K context window is fine for me, and it can keep good proformance. However, 258K is not 400K.
    Could Codex fix this ASAP?

  5. etraut-openai commented on Apr 23, 2026

    @etraut-openai
    Contributor

    The 400K context window includes space for 272K of input + 128K of output. The 258K value is somewhat less than the full 272K to guarantee that compaction is possible. These are the same default values that have been in place for gpt 5.4, gpt 5.3-codex and gpt 5.2.

  6. Cinea4678 commented on Apr 24, 2026

    @Cinea4678

    gpt 5.5 supports 1M context window. Please reopen issue

  7. IlyaShorin commented on Apr 24, 2026

    @IlyaShorin

    The 400K context window includes space for 272K of input + 128K of output. The 258K value is somewhat less than the full 272K to guarantee that compaction is possible. These are the same default values that have been in place for gpt 5.4, gpt 5.3-codex and gpt 5.2.

    Sorry, but why did you closed the issue, if the main problem is that 1m tokens context is gone?

  8. jrenouard commented on Apr 24, 2026

    @jrenouard
    Image
  9. HarryPitsillides commented on Apr 24, 2026

    @HarryPitsillides

    Wtf why is this closed? Is this OpenAI stance now no 1M context? Then why use this vs Claude were 1M is the default?

  10. defteame commented on Apr 24, 2026

    @defteame

    why is this closed? this is clear regression

  11. AlchemistChaos commented on Apr 25, 2026

    @AlchemistChaos

    only see 258k in the codex app... is it 400k or 1M? very unclear

  12. 2-fly-4-ai commented on Apr 25, 2026

    @2-fly-4-ai

    Why is this closed? Mayor regression with no announcement. Wasting everyones time. Must we just guess what's gonna happen?

  13. etraut-openai commented on Apr 25, 2026

    @etraut-openai
    Contributor

    Let's continue the discussion in #19464.

  14. locked and limited conversation to collaborators on Apr 25, 2026
  15. marked gpt 5.5 has a bug #20107 as a duplicate of this issue on Apr 29, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    appIssues related to the Codex desktop appbugSomething isn't workingconfigIssues involving config.toml, config keys, config merging, or config updatescontextIssues related to context management (including compaction)

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions