Auto only waste time.
When Auto Model was
Kimi K2 thinking or gpt 5.1 codex mini low, I get better results but now auto mode in cursor model is unusable.
Hey, thanks for the feedback.
Auto mode should automatically pick the best model for each task. To figure out what’s going wrong, I’ll need a bit more info:
- What exact task are you giving the agent where the result isn’t good?
- Which model is Auto mode choosing right now? (Top-right chat context menu > Copy Request ID and share it)
- What Cursor version are you on? (Help > About Cursor > Copy)
Workaround: try selecting a model manually instead of Auto. For example, Claude Opus 4.5 or GPT-5.1 High for harder tasks, and Gemini 3 Pro or GPT-5.1 Medium for everyday ones. This gives you more control over which model is used.
If you have specific prompt examples where Auto gives a poor result but a specific model does better, please share them. That’ll help the team improve the model selection logic.