3 comments

[ 0.26 ms ] story [ 14.1 ms ] thread
Although this is a good improvement over Qwen3.8 Max but I expected a lot more from Alibaba considering how well other flagship models are performing currently.

Just 26.5% on FrontierCode 2.0 and 48.5% on Terminal V4.

Qwen4 27B might be much more interesting to look out for than Qwen4 Max.

We need a open model which is small and good in coding
the datacentres might go quiet!