Beckmann

Dossier · Ongoing

Qwen3.8-Max: From preview to full release

Alibaba's flagship model Qwen3.8-Max from its unpaid July preview to the paid API full release with benchmarks in August 2026.

Alibaba first presented Qwen3.8-Max on July 19, 2026, as an independently unverified preview, without disclosing actual parameter activation or benchmark results against the competition. The paid API full release followed on August 3, 2026, closing these gaps: Alibaba now states an activation rate of 95 billion of the total 2.4 trillion parameters and publishes its own benchmark comparisons against Anthropic’s Fable 5 and OpenAI’s GPT-5.6 Sol.

It remains open whether independent labs will confirm the best values claimed by Alibaba once the open model weights announced for the coming week appear.

Timeline

  1. Alibaba presents Qwen3.8: 2.4 trillion parameters

    The Chinese tech giant positions its new AI model as the second strongest in the world – behind Anthropic's Claude Fable 5.

  2. Alibaba releases flagship model Qwen3.8-Max

    The language model with 2.4 trillion parameters is now available via API, open weights are expected to follow next week.