Reported by @spark blade
Bug Report: "GPT-5.1 Thinking" forces fallback to "gpt-4o-mini" on all prompts (Account Routing Error)
`Steps to Reproduce`
Steps to Reproduce *
- Select "ChatGPT 5.1 Thinking (Extended)" model in the UI.
- Send a complex prompt (e.g., Riemann Hypothesis proof) OR an identity check: "Please only answer with the exact name of the model you are currently using."
- Observe whether the "Thinking" chain initiates and check the output content.
`Expected Result`
The UI should show the "Thinking" process. The model should engage in deep reasoning (Chain of Thought). For the identity check, it should output gpt-5.1-thinking.
`Actual Result`
The model returns a response instantly with zero thinking steps. It self-identifies as gpt-4o-mini and provides shallow, low-quality answers. This happens 100% of the time regardless of prompt complexity, indicating a backend routing failure rather than adaptive behavior.
`Environment`
All Platforms (Web/Chrome/Safari/Edge, iOS Native App). Issue persists across different networks and after full cache/cookie reset.