KyleHessling1

/

Qwopus3.5-27B-v3-FP8-vllm-ready

like 1

are any of the improvements carrying over when thinking is off, compared to stock 27B? I'm running 27B with thinking off for a specific task where some basic reasoning still helps. Wondering if this model would make sense in that scenario, or would it just be like using vanilla qwen3.5 27B?