It's just been released, and the gguf hasn't been registered yet.
I'm looking forward to it. qwen3.5 27b had really good performance.
I'll test it here and there and write a review.
As for the picture below, don't trust the score too much..
=================
One day of use:
Hardware: rtx 3090 24g, ram: 32g
I was using qwen3.6 35b a3b, but when I try to use 27b, the token output speed degradation is quite significant.
Code was pouring out at 70tok/s, but when using 27b, the code is output at 25tok/s, which is disappointing.
Previously, I was doing work fine with 25tok/s.... but now waiting is...
I haven't compared the results yet, so if there's a definite performance difference, I'll try 27b,
but the results from qwen3.6 35b a3b aren't bad either.. so for now I'll be using qwen3.6 35b a3b.
I need to watch 27b a bit more.
