- Who
- Qwen
- What
- reports benchmark results Qwen4-Max scores 52.69 on the Work360 benchmark, about 10% higher than Qwen3.8-Max
- Quote
The API is live on QwenCloud.
What happened
Qwen announces Qwen4-Max, a model built on a new MoE architecture with native vision and a 1M-token context window. In Qwen’s internal evaluation, it outperforms Qwen3.8-Max using up to about 50% fewer tokens, while average cost per task falls by over 60%. Its API is live on QwenCloud, and Qwen says open weights for smaller Qwen4 models will be released.
Why it matters
Qwen4-Max’s API is live, with Qwen reporting over 60% lower average cost per task than Qwen3.8-Max in its internal evaluation, making the model available to developers at lower stated task costs.
Read next
See the full storyAlibaba Releases Qwen4-Max with Native Vision and Live API Access3 reports
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery!
Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window.
Highlights:
- Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation.
- Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables.
- Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery.
- $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%.
The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder.
Start building today! 🚀
- Blog: qwen.ai/blog?id=qwen4-max
- API: qwencloud.com/models/qwen4-max
- Token Plan: qwencloud.com/pricing/token-p…
More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
The story so far
Alibaba released Qwen4-Max, a 2.4T-parameter MoE model with native vision and a 1M-token context window, with its API now live on QwenCloud and the Qianwen AI platform. The model is also available on Qwen Studio, QwenWork, Qoder, Qwen APP, Alibaba’s Token Plan, OpenRouter, OpenCode Go, Vercel AI Gateway, and Novita Labs. Vercel says its version supports up to 384K output tokens and is priced 25% below Qwen3.8-Max. Qwen lists QwenCloud API pricing at $1.5 per million input tokens, $4.5 per million output tokens, and $0.075 per million tokens for implicit caching, which it says is 70% lower than Qwen3.8-Max. QbitAI reports Qianwen AI platform cache-hit pricing as low as 0.5 yuan per million tokens. Alibaba says open weights for smaller Qwen4 models will be released over time. In Qwen’s evaluations, Qwen4-Max scored 52.69 on Work360, about 10% above Qwen3.8-Max, while cutting token use by about 24% at Xhigh and 35% at High and lowering average task cost from $4.43 to $1.67. On Terminal-Bench 4.0, Qwen says token use fell by about 48% and 49% at those settings, while QbitAI reports a score of 48.5 versus 38.9 for Qwen3.8-Max, with average task cost down 58%.
How the story developed · 2
View full event- 11/16 22:39 (UTC+8)Alibaba Releases Qwen4-Max with 2.4T Parameters and 1M-Token ContextQbitAI
- 11/16 10:38 (UTC+8)
Qwen4-Max launches with native vision and a 1M-token context window This report
