Alibaba announces Qwen3.8-Max flagship model with 2.4 trillion parameters
Alibaba announced Qwen3.8-Max, a flagship model designed for long-horizon tasks such as reproducing research papers and designing chips autonomously. The team plans to release the weights next week.
First seen 3 Aug, 10:48 UTC on The Decoder7 sourcesLast update 73d ago
AlibabaLaunch · 3 Aug
Qwen3.8 Max
Frontier-class model launch
$2.00 / $6.00per M tokens, in / out
Openweights
How Qwen3.8 Max compares
List prices per million tokens, context and Arena's text leaderboard, against Alibaba's other top models and the leading labs' best. The percentages are Qwen3.8 Max's price against each.
| Model | Released | Input | Output | Context | Arena |
|---|---|---|---|---|---|
| Qwen3.8 Max | September 2026 | $2.00 | $6.00 | 1M | not ranked yet |
| Qwen3.8 Flash | August 2026 | $0.15 +1233% | $0.47 +1177% | 1M | not ranked yet |
| Qwen3.8 27B | August 2026 | $0.42 +376% | $2.55 +135% | 1M | not ranked yet |
| Claude Opus 5.5 | September 2026 | $4.00 -50% | $20.00 -70% | 1M | #2 1507 |
| GPT-6 Astra | September 2026 | $10.00 -80% | $50.00 -88% | 1.05M | #30 1475 |
| Gemini 4 Argon | October 2026 | $2.00 | $10.00 -40% | 1M | #1 1525 |
Arena ranks the best effort setting of each model among all models on its text leaderboard. Every model and price.
What the sources say
Linked, never rewritten. Official means the lab itself.
Official3
Hugging Face · Qwen· 5 Aug, 08:22Qwen/Qwen3.8-27B published on Hugging FaceHugging Face · Qwen· 8 Aug, 01:49Qwen/Qwen3.8-2.4T-A95B-FP8 published on Hugging FaceHugging Face · Qwen· 8 Aug, 01:50Qwen/Qwen3.8-2.4T-A95B published on Hugging FacePress4
The Decoder· 3 Aug, 10:48Alibaba's open-weight Qwen3.8-Max takes on long-horizon AI tasks with 2.4 trillion parametersThe Decoder· 3 Aug, 17:12Alibaba's new Qwen model is also taking your job, but this time it's greatSiliconANGLE· 3 Aug, 20:14Alibaba debuts Qwen3.8-Max model with 2.4T parametersThe Decoder· 6 Aug, 13:33Qwen3.8 Max catches Claude Opus 4.8 but Kimi K3 still scores higher for 25 percent lessHow it unfolded
Every source in the order it appeared. Times in UTC.
3 Aug 10:48The Decoder Alibaba's open-weight Qwen3.8-Max takes on long-horizon AI tasks with 2.4 trillion parameters3 Aug 17:12The Decoder Alibaba's new Qwen model is also taking your job, but this time it's great3 Aug 20:14SiliconANGLE Alibaba debuts Qwen3.8-Max model with 2.4T parameters5 Aug 08:22Hugging Face · Qwen Qwen/Qwen3.8-27B published on Hugging Face6 Aug 13:33The Decoder Qwen3.8 Max catches Claude Opus 4.8 but Kimi K3 still scores higher for 25 percent less8 Aug 01:49Hugging Face · Qwen Qwen/Qwen3.8-2.4T-A95B-FP8 published on Hugging Face8 Aug 01:50Hugging Face · Qwen Qwen/Qwen3.8-2.4T-A95B published on Hugging Face
Keep up with Alibaba
Alibaba on the ScoreIts timelineFollow to mark Alibaba's stories across the site. No account needed.