← All stories
● Covered by 1 source · 1 reportMedium impact1 neutral

Alibaba Releases Qwen4-Max, a Proprietary AI Model for Complex Professional Tasks

🔄 Updated 5d ago
New to BrevFeed? We gather this story from every outlet covering it into one summary — ranked by real-world impact, not just the latest headline — so you never miss what matters. What is BrevFeed? →

Key points

  • Alibaba released Qwen4-Max, focusing on professional tasks and AI agents.
  • The model uses a MoE architecture with 2.4 trillion backbone parameters.
  • Qwen4-Max is proprietary, unlike previous open-weight Qwen releases.
  • It supports text and images, with a one-million-token context window.

Qwen4-Max Unveiled

Alibaba has officially introduced Qwen4-Max, the initial model within its new Qwen4 series. This model is specifically developed for complex professional tasks and autonomous AI agents. The company previously hinted at the Qwen4 family, promising enhancements in coding, reasoning, and real-world knowledge.

Proprietary Release Strategy

Unlike previous Qwen models, which were released with open weights, Alibaba is keeping Qwen4-Max proprietary. This marks a shift in strategy for the company's most powerful AI model, despite its earlier contributions to the open-source AI community.

Architectural Details and Capabilities

Qwen4-Max is built on a new mixture-of-experts (MoE) architecture, featuring 2.4 trillion backbone parameters and activating 71 billion parameters per token. It also incorporates 328 billion parameters for N-gram Embedding to improve text pattern processing. The model supports both image and text inputs and has a one-million-token context window, allowing it to handle large volumes of information in a single interaction.

Availability and Performance

Qwen4-Max is accessible via Alibaba’s QwenCloud platform, with API pricing starting at $1.50 per million input tokens and $4.50 per million output tokens. It is also available through QwenWork, Qoder, and QwenStudio. Vercel has integrated Qwen4-Max into its AI Gateway, enabling its use with coding tools like Claude Code, Codex, and Cursor.

Alibaba reports that Qwen4-Max shows improvements over its predecessor, Qwen3.8-Max, particularly in complex tasks involving multiple tools and autonomous execution. On the internal Work360 benchmark, Qwen4-Max scored 52.69 compared to Qwen3.8-Max's 47.74, and demonstrated 24% to 35% fewer token usages. Token savings were even greater on Terminal-Bench 4.0, a test for AI agents in terminal environments, where usage nearly halved.

✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →

The daily brief

One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.

One email a day. Unsubscribe in one click, any time.

Today's brief

Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.

~4 min · 3 stories · Nov 21

▶ Play today's brief Listen on Spotify

New every morning, and the back catalogue is archived by date.

Reporting from

Alibaba has launched Qwen4-Max, the first model in its Qwen4 family, designed for complex professional tasks and autonomous AI agents. This new model, built on a mixture-of-experts (MoE) architecture, is being released as proprietary, departing from Alibaba's previous open-weight Qwen models.