Qwen3.8-27B released! (Qwen compares it again Opus4.6 Max)

[third eye]

huggingface.co/Qwen/Qwen3.8-27B

Crazy benchmark scores, if not benchmaxxed.

Several posts and discussion on reddit.com/r/LocalLLaMA.

Most will be interested in running the quants, not the full weights: unsloth.ai/docs/models/qwen3.8#quantization-analysis

artificialanalysis.ai will evaluate it soon.

[third eye]

against*