API currently does not allow reasoning, as we only have llama models, haven't heard much pull from Groq GPT-OSS 120b, so we haven't considered implementing yet. Currently slated are azure 5.1, opus 4.5, and gemini 3.