Benchmarks

3 articles tagged with this topic

Qwen3.8 Hits 94% on Mac — But Benchmarks Are Breaking Down

Alibaba quietly released Qwen3.8; a developer hit 94% on a ~$7,000 Mac. More telling: the blogger admits "models are getting too good to differentiate

2d ago2 min read

QwenAlibaba

Qwen 3 'Goated' at Low/Mid Compute — Alibaba Silences Overthinking Critics

Qwen 3 crushes benchmarks at low/mid compute, debunking the "overthinking" myth. The open-source ecosystem just got stronger.

Aug 212 min read

ZhipuGLM-5

GLM-5.3 Hits Artificial Analysis — China's First Open-Source Flagship Vetted

Zhipu's GLM-5.3 completes Artificial Analysis benchmarks — first Chinese open-source model to land in the global top tier with an independent score.

Aug 192 min read