#alibaba

At the time of writing there is no Wan 3.0: no Alibaba announcement, no weights, no listing. July's real output was one open-weight model and two research papers, not three.

Qwen3.8-Max ships as a 2.4T-parameter MoE with 95B active and a 1M context. What the vendor benchmark table shows, and what is still only promised.

Alibaba runs two parallel video lines: audio-native HappyHorse and open-leaning Wan 2.7. Neither is new, and their arena snapshots do not cleanly reconcile.

Qwen-Image-3.0 generates dense single-pass layouts with fine text, but Alibaba shipped no weights, benchmarks, or license. Verify output before production use.

Alibaba previewed Qwen3.8-Max at WAIC Shanghai, claiming 2.4 trillion parameters and second only to Fable 5 — yet shipped zero benchmarks to back it.

Anthropic told US senators that Alibaba ran 28.8M Claude queries via 25,000 fake accounts, the largest model-distillation campaign it has disclosed.

Alibaba releases Qwen 3.5 small series from 0.8B to 9B parameters. The 9B model beats GPT-class models on GPQA Diamond benchmark for on-device AI deployment.

Qwen 3.5-397B scores 83.6 on LiveCodeBench v6 and 91.3 on AIME26 with 17B active MoE params. Benchmarks vs GPT-5.2, Claude, and pricing details.

Alibaba's Qwen3-Max-Thinking scores 58.3 on HLE and 100% on AIME25. Test-time scaling, adaptive tools, and competitive $1.20/M token pricing.

Master Alibaba's Qwen App Beta with deep research, AI coding, voice calls, and AI camera. Complete guide to China's AI super app strategy and ecosystem.