I know it seems a strange request but I would be interested in getting access to o3-mini low and medium.
I find for anything other than “one-shot” requests o3-mini spends too long thinking.
I like the back and forth chatting with Claude Sonnet that you just don’t get with o3-mini-high. You really have to get your prompt perfect and sit waiting to see what you forgot to mention.
In theory low or medium could be smarter than Sonnet with a similar latency but with faster token generation.