GPT-6 consistently returns “Invalid prompt” across conversations and clients, while other models work

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

[Enter the exact version shown in “About Codex”]

What subscription do you have?

[Enter the subscription associated with the affected account, e.g. ChatGPT Pro]

What platform is your computer?

[Operating system, version, and architecture]

What issue are you seeing?

Summary

I am unable to use GPT-6 in any of the conversations or clients I have tried. Requests to GPT-6 consistently fail with the “Invalid prompt” error below, while other models remain usable.

Starting a different conversation or switching to another client does not resolve the issue. The failure appears to be specific to GPT-6 in my experience, rather than limited to a single conversation or client.

Error message

Invalid prompt: your prompt was flagged as potentially violating our usage policy. Please try again with a different prompt:

The error also points to:
https://platform.openai.com/docs/guides/reasoning#advice-on-prompting

Steps to reproduce

1. Open a conversation in an affected client and select GPT-6.
2. Submit a prompt.
3. Observe the “Invalid prompt” error instead of a response.
4. Try GPT-6 in another conversation or another client. The error persists.
5. Select another available model. Other models remain usable.

Expected behavior

GPT-6 should respond to permitted requests normally. If access is unavailable for another reason, the error should clearly identify that reason and provide actionable guidance.

Actual behavior

GPT-6 repeatedly rejects requests with a message saying the prompt was flagged as potentially violating the usage policy. Changing conversations or clients does not restore access.

What I have tried

• Using different conversations.
• Switching between clients.
• Using other models to check whether the problem affects all models.

The first two steps did not resolve the GPT-6 issue. Other models continue to work.

Impact

This is blocking my ongoing mathematical modeling and coding work. I specifically need GPT-6 for this work, so using another model does not fully address the problem.

Additional details

• Exact model name displayed in the interface: [Enter model name]
• Clients tested and their versions: [Enter details]
• Approximate time the issue occurred, including time zone: [Enter details]
• Example prompt that actually failed, with sensitive information removed: [Enter prompt]

Could you please investigate why GPT-6 requests are consistently rejected across the clients and conversations I have tried, while other models remain usable? Please let me know what diagnostic information would help identify the cause and restore normal access.

What steps can reproduce the bug?

Summary

I am unable to use GPT-6 in any of the conversations or clients I have tried. Requests to GPT-6 consistently fail with the “Invalid prompt” error below, while other models remain usable.

Starting a different conversation or switching to another client does not resolve the issue. The failure appears to be specific to GPT-6 in my experience, rather than limited to a single conversation or client.

Error message

Invalid prompt: your prompt was flagged as potentially violating our usage policy. Please try again with a different prompt:

The error also points to:
https://platform.openai.com/docs/guides/reasoning#advice-on-prompting

Steps to reproduce

1. Open a conversation in an affected client and select GPT-6.
2. Submit a prompt.
3. Observe the “Invalid prompt” error instead of a response.
4. Try GPT-6 in another conversation or another client. The error persists.
5. Select another available model. Other models remain usable.

Expected behavior

GPT-6 should respond to permitted requests normally. If access is unavailable for another reason, the error should clearly identify that reason and provide actionable guidance.

Actual behavior

GPT-6 repeatedly rejects requests with a message saying the prompt was flagged as potentially violating the usage policy. Changing conversations or clients does not restore access.

What I have tried

• Using different conversations.
• Switching between clients.
• Using other models to check whether the problem affects all models.

The first two steps did not resolve the GPT-6 issue. Other models continue to work.

Impact

This is blocking my ongoing mathematical modeling and coding work. I specifically need GPT-6 for this work, so using another model does not fully address the problem.

Additional details

• Exact model name displayed in the interface: [Enter model name]
• Clients tested and their versions: [Enter details]
• Approximate time the issue occurred, including time zone: [Enter details]
• Example prompt that actually failed, with sensitive information removed: [Enter prompt]

Could you please investigate why GPT-6 requests are consistently rejected across the clients and conversations I have tried, while other models remain usable? Please let me know what diagnostic information would help identify the cause and restore normal access.

What is the expected behavior?

No response

Additional information

01a08d2f-de55-7823-9159-a4e530a5417e

The report names no repository file, test, or entry point, and leaves the model name, client versions, timing, and failed prompt as placeholders. Start by checking whether the behavior is reproducible with complete diagnostic details; done would require a confirmed cause and restored GPT-6 access without affecting models that already work.