The Agent seems to behave much differently when preparing context for o4-mini and o3, as compared to Sonnet 3.7. Sonnet 3.7 (thinking) does quick visible thinking with the context I provided, and gets immediately to work, sometimes with no initial tool calls. With o4-mini and o3, the agent makes 1…

Apart from the questions I raised above, this feature request would help the situation: [image] Re-use of context assembled by Agent Feature Requests For some models (o4-mini, o3, for example), Agent spends a lot of time building up context before actually running the big …

Agent behavior much different for o4-mini, o3

Ideas Feedback

121kr0x33 (0x33agent) May 1, 2025, 5:57pm 2

Same here, o4-mini is painfully slow, but it’s more accurate than gemini 2.5 pro in my case

Cursor is too slow or doesn't generate responses on paid plan

Topic		Replies	Views
Re-use of context assembled by Agent Feature Requests	2	92	May 15, 2025
Add reasoning models to agent mode Feature Requests	13	746	February 3, 2025
O3-mini not agentic? Bug Reports	32	3637	February 22, 2025
O3-mini in agent mode fails function calls (credits dump) Bug Reports	8	630	February 16, 2025
Comparison of models / cursor vs native / agent vs normal / Discussions	1	685	February 16, 2025

Agent behavior much different for o4-mini, o3

Related topics