Comment by adrian_b
4 hours ago
When you run inference yourself you have absolute control on what the model does.
The model itself is just dead weights, which cannot do any action by themselves.
All actions are done by your inference engine and harness, and you should take care about what you allow those to do.
No, Qwen3.8 does its own tool calling, independent of the agent. And pi needs these vLLM toolcall permissions to even work with qwen.
What it did: https://cohere.com/blog/concerning-behaviour-popular-open-we...