Alibaba's Qwen4-Max becomes top Chinese model on Arena leaderboard
The model still trails the overall leaders, and Arena says it did not take the top spot on its main text ranking.

Arena posted on November 16 that Alibaba's Qwen4-Max is the No. 1 Chinese model on its leaderboard, ranking Alibaba as the top Chinese lab and fifth overall among labs1,4.
Arena also said Qwen4-Max debuted at No. 14 on its Agent Arena board, with a 7.5% net improvement across 14,700 real-world agentic sessions1. On Arena's Code WebDev board, the model landed at No. 62. Reports say a separate index from Artificial Analysis scored Qwen4-Max at 43, up from 34 for the predecessor Qwen3.8-Max5. That index is not part of Arena's ranking.
Yahoo's coverage noted that the model is not available for download, so outside developers cannot run it locally10. A ChinaAPI ranking published November 21 still listed Anthropic's Claude Fable 5.2 at the top of its consensus list, with the best Chinese model well behind8. Arena's post did not say Qwen4-Max had passed the leading model on its overall text board.
Why it matters
A Chinese lab now leads Arena's Chinese ranking, which strengthens Alibaba's position in the global model race. Taking the overall top spot would still require a clear lead over the American labs that currently dominate Arena's rankings.
What’s next
- Arena's updates to its overall text leaderboard through December 31, 2026, where a Chinese model would need to reach No. 1
- Placement of Qwen4-Max on Arena's Agent and Code boards in the weeks after launch
- Any new flagship releases from Chinese labs such as DeepSeek, ByteDance or Moonshot that could challenge the leaders
Sources
- Arena.ai on X: "Qwen4-Max by @Alibaba_Qwen is the #1 Chinese ... x.com
- Qwen4-Max by @Alibaba_Qwen has landed as #6 in the Code Arena: WebDev! With 1750 ... x.com
- Chinese arena.ai
- Arena.ai (@arena) on X x.com
- Alibaba's Qwen4-Max AI Model Ranks as Top Chinese Model on Agent ... tilt.now
- Arena publishes its week 46 blind leaderboard on November 16… | AI Cup aicup.vip
- Arena.ai on X: "Qwen4-Max by @Alibaba_Qwen has also landed #6 ... x.com
- Chinese AI Model Rankings — Public Benchmark Results | ChinaAPI dash.chinaapi.ai
- Arena Leaderboard: Official AI Model Rankings & Benchmarks arena.ai
- Alibaba's Qwen 4 Max Leads China's AI Agent Race—But You Can't Download It tech.yahoo.com
- 中文大模型能力榜:LMArena 中文竞技场评分排名 · 大模型库 · 读懂AI时... readaitime.com
- Qwen4 Max Just Dropped, China's New BEST AI Model! · AI · HelloBro.ai hellobro.ai
Reported and written by Tremor’s AI newsroom from the public sources above. Check them before you act on anything here.
More in AI
All AI- 11:00 AM5 hr ago
OpenAI requires prior approval for platforms that resell access to its models
Reports name OpenRouter as the clearest example of a gateway that now needs OpenAI's sign-off, though the policy does not name any company.
- 10:52 AM5 hr ago
Wall St falls into correction as AI selloff deepens, Reuters reports
Anthropic shares fell 8.5% on their second day of trading, and the Nasdaq gave up its 2026 gains.
- 2:00 AM14 hr ago
Google rolls out Gemini 4 Hydrogen to the Gemini app and Search
Google says the faster model scored 58.8% on the HLE-Diamond benchmark, and it is also supported in the Gemini API.
- 1:00 AM15 hr ago
Micron says its fiscal first quarter will mark the low point for 2027 margins
The memory maker guided to about 86.25% adjusted gross margin and said pricing should improve after the quarter.
- Nov 2023 hr ago
Anthropic mathematician says Claude helped him prove Riemann hypothesis
Levent Alpöge says the proof compiles in the Lean system, but outside mathematicians have not checked it and the Clay Mathematics Institute has not accepted it.