Comment by Patrick_Devine

Ollama only uses llamacpp for running legacy models. gpt-oss runs entirely in the ollama engine.

You don't need to use Turbo mode; it's just there for people who don't have capable enough GPUs.