Qwen 3.8-Max
Qwen 3.8-Max
Alibaba's flagship, and since August its weights are published too.
Updated
The read
3.8-Max is the top of Alibaba's line, and in August the company reversed course and published its weights as a 2.4-trillion-parameter sparse model. It sits in the top group on both independent leaderboards, covers more languages than almost anything else here, and a stable API snapshot from September ended the preview churn. The licence is permissive for most users but requires a separate agreement from large companies selling it as a service.
Where it shines
- Broadest language coverage of any frontier model
- Weights published as a 2.4-trillion-parameter sparse model
- API priced well below Western flagships
Worth knowing first
- Large companies selling it as a service need a separate licence
- The open weights are text only and always reason before answering
Index score
85.1
Provisional. This score is read from the product's public capability, not from the one-prompt rebuild the fully reviewed entries went through, so treat it as a placing rather than a verdict.
- Rank
- 18 of 73
- Builds
- LLMs
- Category
- Frontier
- Output
- Published weights under its own licence, plus an API
- Pricing
- Free tier
- Best for
- Multilingual frontier work, hosted or on your own hardware
Scores are our own editorial judgement, weighted craft 55, speed 10, control 15 and value 20.
Compare it with
Also worth a look
Full index- 01Claude Opus 5.5Claude Opus 5.5ProvisionalFable-level answers for less than the Opus it replaces. The new default.Frontier modelPaid only92.8
- 03Gemini 4 ArgonGemini 4 ArgonProvisionalTop of the human-preference leaderboard, and almost nobody can use it yet.Frontier modelPaid only90.6
- 04GPT-6.1 SolGPT-6.1 SolProvisionalNear-Astra answers at a fifth of the price. The OpenAI model most teams should deploy now.Frontier modelPaid only90.3
