99% of AI agents just write text.
Qwen 3.8 Max actually runs your business. Alibaba just dropped a massive new model. It scored an 86.1 on OSWorld-Verified. That means it can click apps and finish real work. It beats Fable 5 and GPT-5.6 Sol Max. I plugged it into our Agent
Gemini 3.7 Flash vs Qwen 3.8 vs Grok 4.6 vs GLM 5.3
's Qwen 3.8 27B NVFP4 did this with 240k tokens on NVIDIAAI 's DGX Spark in Opencode
Single html file game "WolfQwenstein 3.8 27B". Two weapons, 3 maps, multiple enemies. Even has sound and music but couldn't record it on my Mac. Alibaba_Qwen
Qwen 3.8 27B, self-hosted on a single rented GPU
Builds a complete three.js scene from a prompt: sunken Atlantis with temple, columns, schools of fish and caustics, plus a scripted 20-second camera flight. Same prompt as Claude Opus 4.6,
You have reached the end of the archive
All of qwen38