I think you could reconsider DeepSeek-R1: it's actually really good. In comparison, o3-mini gets very vague in its reasoning, and gives surprisingly unhelpful

I think you could reconsider DeepSeek-R1: it's actually really good. In comparison, o3-mini gets very vague in its reasoning, and gives surprisingly unhelpful answers (getting too short). Plus, let's not forget, R1 is available to use and modify under MIT license, which is great.