The video was funny. It appeared to think and rethink through the same thought process ~7 times. Seemed like overkill.
Great addition regardless. Good to see open weight model progress.
Was just wanting something like this today, running qwen3 q2 model. Spent 3 paragraphs working out how to respond to "hello how are you".
The third options would be a constrained thinking, ie 1-2 sentences max before output.
The video was funny. It appeared to think and rethink through the same thought process ~7 times. Seemed like overkill.
Great addition regardless. Good to see open weight model progress.