[daemonologist]

There is no "coder" version of Qwen 3.6; I think they just mean it's a coding-focused model of similar size and performance (to Qwen 3.6 35B-A3B).

Regular Qwen 3.6 benchmarks slightly better and has much wider software support though, so this is probably of interest only to organizations which disallow models trained in China.

[kadoban]

I mean, Qwen 3.6 kicks ass. I don't know who these people are, but if their first outing is "not quite as good as Qwen 3.6", that's not a bad start by any means.

30B vs 35B isn't nothing either.

If it ends up just being some tweaks to someone else's weights, then meh.

[mtone]

It was trained from scratch by Cohere. They're the only Canadian AI lab - I'm glad they're releasing open weights and I wish them luck catching up!