I ran Qwen3.8-27B against Opus, Sonnet, GPT and others. Results inside.

Article automatically generated from technical news.

I created a small testing rig to evaluate new open source models as they drop, and with the much anticipated release of Qwen3.8-27B, I was eager to see how it cross-compares with frontier and strong local models. The rig My test rig is an M5 Max MacBook Pro, 128GB. Locally I ran Qwen3.8-27B on xhigh and medium thinking modes via LM Studio, MLX 8-bit, temp 1.0 / top_p 0.95 / top_k 20, context 131,072 and DeepSeek V4 Flash "0731" 2-bit-imatrix q2-q4, served by antire