Do you realize yet that Qwen 3.6 27B is the BEST local coding model in the world that can fit in under 100 GB of local VRAM? And it's only a fraction of that size. Alibaba hit this one out of the park! I'm coding with it today as a test, and so far it's super impressive.
To all those people concerned about how much AI hallucinates, have you checked the hallucination rate of HUMANS lately? Ask almost any question, about politics, religion, history, physics, medicine, vaccines, etc., and you'll find that the human hallucination rate is atrociously bad.