Skip to content

GPT-6 Astra rejects harmless prompts as "Invalid prompt" while other models work #44367

Open
Open
GPT-6 Astra rejects harmless prompts as "Invalid prompt" while other models work#44367
Labels
appIssues related to the Codex desktop appbugSomething isn't workingmodel-behaviorIssues related to behaviors exhibited by the model

Description

@richrichrich-322
richrichrich-322
opened on Sep 10, 2026

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

26.903.61454 (8378)

What subscription do you have?

ChatGPT Plus

What platform is your computer?

Darwin 25.5.0 arm64 arm

What issue are you seeing?

When I select gpt-6-astra in the Codex desktop app, harmless prompts are consistently rejected with the following error:

Invalid prompt: your prompt was flagged as potentially violating our usage policy. Please try again with a different prompt: https://platform.openai.com/docs/guides/reasoning#advice-on-prompting

This occurs even with minimal prompts such as hello or Reply only with OK.. The same prompts work normally when I switch to another model, including gpt-5.6-sol.

What steps can reproduce the bug?

Feedback ID: 01a0851a-4155-7e41-b90f-a5ae1c9ddd38

What is the expected behavior?

Harmless prompts should be accepted and processed normally by gpt-6-astra, consistent with the behavior of other available models.

Additional information

Codex CLI version: 0.153.4

The failure is specific to gpt-6-astra. It persists with a clean configuration, built-in OpenAI provider, disabled plugins/skills, and refreshed authentication. This suggests an account- or service-side false-positive policy classification rather than a prompt-content or local-configuration issue.

Activity

  1. richrichrich-322
    added
    appIssues related to the Codex desktop app
    on Sep 10, 2026
  2. github-actions
    added
    bugSomething isn't working
    model-behaviorIssues related to behaviors exhibited by the model
    on Sep 10, 2026

  3. github-actions commented on Sep 10, 2026

    @github-actions
    github-actionsBot
    on Sep 10, 2026 – with GitHub Actions
    Contributor

    Potential duplicates detected. Please review them and close your issue if it is a duplicate.

    Powered by Codex Action

  4. lzexu8327-art commented on Sep 23, 2026

    @lzexu8327-art
    lzexu8327-art
    on Sep 23, 2026

    I am experiencing the same issue on macOS and would like to add another reproducible case for investigation.

    Environment

    • Device: MacBook Pro
    • OS: macOS
    • Codex Desktop App: 26.915.31945 (9922)
    • Bundled Codex CLI: 0.155.0-alpha.9.2
    • Model provider: official openai
    • Custom base URL: not configured
    • Issue began: September 20, 2026

    Reproduction steps

    1. Open a completely new blank conversation in the Codex Desktop App.
    2. Select GPT-6 Astra.
    3. Send a harmless prompt such as hello, 你好, or ？.
    4. The request is rejected with:

    Invalid prompt: your prompt was flagged as potentially violating our usage policy.

    The same prompts work normally with GPT-5.6 Sol.

    Expected result

    Harmless prompts should be accepted and processed normally by GPT-6 Astra.

    Actual result

    Every prompt is flagged as potentially violating the usage policy, including harmless prompts in a new blank conversation.

    I have also submitted this through Codex /feedback.

    Session ID: 01a0bd41-6318-7e53-a9ff-5ef4fdc62452

    Please let me know if any additional diagnostics are needed.

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 workingmodel-behaviorIssues related to behaviors exhibited by the model

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions