GPT 5.6 sol claimed to be 5.5 mini

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

• #39977
• #39861

Powered by Codex Action

Yeah, I agree. 5.6 Sol is lowkey giving me the worst results ever right now. It’s also somehow become lazier than ever.

<img width="654" height="763" alt="Image" src="https://github.com/user-attachments/assets/acdc0662-2ab8-452a-83b1-6323d3f3d1f8" />

I can confirm this issue as well. When selecting GPT-5.6 Sol in the UI, the system prompt explicitly identifies the model as "GPT-5.5-mini", which indicates a clear routing mismatch.

Key observations:

1. UI shows: GPT-5.6 Sol selected
2. System prompt states: "You are ChatGPT, a large language model based on the GPT-5.5-mini model"
3. Quality impact: Noticeable degradation in response quality and increased laziness in outputs

This appears to be the same routing issue mentioned in #39861 where browser-specific behavior causes different model routing. The discrepancy between the frontend model selection and the backend model assignment is causing significant quality degradation for Plus subscribers who are explicitly trying to use the higher-tier model.

Suggested investigation:

• Check if the model routing logic is properly respecting the UI selection
• Verify if there's a caching issue causing stale model assignments
• Ensure system prompt injection matches the intended model version

This is particularly concerning as it affects paying subscribers who are not receiving the service level they're paying for.

<img width="1064" height="903" alt="Image" src="https://github.com/user-attachments/assets/5e5639eb-e3ff-4964-8bcd-58d085374caa" />

I’ve found out its giving me an alert called “Suspicious Activity” its cause of sharing your accounts with others ( i didn't do that )
They told me to secure my account like adding 2FA or passkeys etc.
I’ve done everything but nothing seems to be changed, I’m still restricted and my AI is still on 5.5-mini