Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

Re:Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model by harkh on Wednesday April 22, 2026 @05:24PM (#47869474)

I agree, what would you say are currently the best local options?

[ Parent | on HN ]

Re:Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model by zargon on Wednesday April 22, 2026 @08:54PM (#47871144)

For FIM, there's Qwen3 Coder Next.

Although Mistral's model card seems to indicate that Devstral 2 doesn't support FIM, it seems very odd that it wouldn't. I have been meaning to test it.

[ Parent | on HN ]

Re:Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model by thot_experiment on Wednesday April 22, 2026 @10:45PM (#47871772)

Qwen Coder 30B A3B is far better than Qwen Coder Next imo. I may have inference issues or it's just a problem with running Coder Next at IQ4 XS, vs Q8 for the earlier/smaller model but I don't find the 80B to be much better at coding, even in instruct mode, and the insane speed and low latency of the smaller model is way more useful. Good one-line completions often happen in 300ms.

[ Parent | on HN ]