Comment by unfirehose

this is a good model. I benchmark reasoned answers to qwen 3.6 27b (no think)+ bash and it held up.