Alibaba releases Qwen4-Max, priced below Opus 5.5, GPT-6 and Kimi
Original title: Alibaba's Qwen4-Max takes on Opus 5.5, GPT-6, and Kimi K3.1 Pro at much lower prices
The 2.4-trillion-parameter mixture-of-experts model activates 71 billion parameters per token and offers a one-million-token context window.
Alibaba released Qwen4-Max as its flagship for professional work in industry, science, and finance. A hybrid design compresses prior context across three layers before a fourth layer attends to selected parts, and an N-gram embedding layer has 328 billion parameters. Alibaba says Work360 task cost fell 62 percent, from $4.43 to $1.67, while the Artificial Analysis Intelligence Index score rose nine points to 43. Its API lists $1.50 per million input tokens and $4.50 per million output tokens.
Unconfirmed: Cost, benchmark, and partner-deployment figures are claims by Alibaba or Artificial Analysis, not independently verified here.
- Qwen4-Max-Base improved 5 to 13 percent over Qwen3.8-Max-Base with 25 percent fewer active parameters.
- It scored 46.7 on Agents' Last Exam v1.1, the best result in that comparison.
- Artificial Analysis prices one Intelligence Index task at $3.27 versus $10.02 for Qwen3.8-Max.
- Cache hits cost $0.075, and output use is about 105,000 tokens per benchmark task.
- Alibaba open-sourced Qwen-Live-Delivery-Plugin so users can revise long-running agent work early.
- Vendors
- Alibaba
- Organizations
- Artificial Analysis
- AI models
- Qwen4-MaxOpus 5.5GPT-6Kimi K3.1 Pro
From the original article
7 min read at the-decoder.com
Alibaba launches Qwen4-Max, a new flagship model with 2. 4 trillion parameters that activates only 71 billion per token. A hybrid architecture sharply cuts costs on long contexts.
By Jonathan Kemper · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.