Qwen3.8-Max says it beats GPT-5.6 and Fable 5 at computer use. Here's my checklist before I believe any open-weights release

Qwen3.8-Max says it beats GPT-5.6 and Fable 5 at computer use. Here's my checklist before I believe any…

Alibaba's Qwen3.8-Max claims 86.1 OSWorld agentic, beating GPT-5.6 Sol and Claude Fable 5; 27B open-weights promised. If license clears and 27B holds, $0.25/M cached tokens reshape agent economics—but wait for independent benchmarks before production moves.