Users of ChatGPT have reported that the platform intentionally switches the selected model to GPT-6, even when a different model like GPT-5.6 High is explicitly chosen. This behavior occurs when creating new chats, where the model selection appears to be GPT-5.6 High, but the underlying request is rerouted to GPT-6 at a medium thinking level. The exact nature of GPT-6 remains unknown, with speculation suggesting it might be related to the Luna model. AI
IMPACT Potential user confusion and unexpected performance characteristics due to automatic model switching in ChatGPT.
RANK_REASON User report of unexpected model behavior in a widely used product.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →