We use o3 high on the Open AI forum (only visible to tl3) and it is really stellar, if people can afford a reasoning model, it is a very interesting choice. Gemini 2.5 pro has also provided very interesting results.
2 Likes