Currently (as in v. 0.3.3), "thinking" is a boolean flag. However, GPT-OSS "thinking" support is not a boolean value, but: "low", "medium", or "high" (per Ollama's docs.)
If you think this is a good idea, I'll be more than happy to try to implement this, under your guidance.
Many thanks,
-Yvonne