What version of 03 mini is being used by cursor ? Is it the max version or min version? @danperks @jake @truell20
There are three different levels of reasoning—low, medium, and high.
I believe the current o3-mini is using the medium reasoning effort.
hey jake,
can you give any additional context that leads you to think medium … some users are thinking it’s low, and everyone (understandably) is hoping we can get high. 
I asked @ericzakariasson, who works at Cursor 
I believe Cursor are starting with medium to give Cursor users access to o3-mini as quickly as possible, and are actively working on evaluating high and improving the system prompts (since my my comment 9h ago, they have already improved the system prompt for o3-mini).
Does a single o3-mini prompt with the high model count as a premium model fast request or does it use usage based pricing like o1? According to OpenAI pricing, price per 1m tokens is $2.50/$10.00 I/O for GPT-4o which is a premium model while o3-mini is $1.10/$4.40, but it generates more reasoning tokens with high.
According to our docs (Cursor – Models), o3-mini counts as 1/3 of a fast premium request and costs $0.01 per request. The model always has high reasoning as set by our backend!
The token generation and pricing isn’t directly comparable to GPT-4o since they use different systems, but o3-mini is designed to be more cost-effective while still delivering good results