Alibaba launches Qwen4-Max with 1M-token context


Alibaba’s new flagship model is positioned for coding, game development and agent tasks, with a 1 million token context window and lower pricing than top US competitors.
Alibaba has introduced Qwen4-Max, its most powerful flagship AI model, featuring a one-million-token context window.
What the new model can do:
- It surpasses Opus 5.5 and Kimi K3.1 Pro in programming. On the Agents’ Last Exam benchmark, it scored 46.7%, the best result among all models.
- It is suitable for game development and integrates easily with Unity and Unreal Engine.
- It is significantly cheaper than top American competitors: $1.5 per million input tokens and $4.5 per million output tokens.
- It uses 50% fewer tokens, while task execution costs have dropped by 62% compared with the previous version.
- The model is already available in the Qwen app and through the API.
Why it matters
AnalysisThe launch could intensify competition in the LLM market, especially for developers and studios looking for high-performance coding and agent models at lower cost.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Alibaba Qwen (Leader in Open Coding and Mathematics)
A series of high-performance open models from Alibaba's cloud division (Qwen 2.5, Qwen Coder). Recognized as a leader among open models in coding, mathematics, and multilingual text processing.
Complete Guide to Higgsfield Cinema Studio 4.0: Scene Management, Soul 2.0, and API Automation
In-depth practical breakdown of the major Higgsfield update: Cinema Studio 4.0, consistent characters with Soul 2.0, API-based generation, and a ready-to-use UGC pipeline.
Part I. AI Agent Architecture: The Foundations
The eleven foundational building blocks of modern autonomous agents: model taxonomy, the agentic runtime loop, advanced prompting, distillation, skills architecture, memory systems, function calling, and FinOps.
Function Calling / Tool Calling
A technical protocol and standard for LLM interaction with external software. Instead of free text, the model returns valid JSON containing the function name and typed arguments according to JSON Schema, enabling the backend to deterministically execute actions in real APIs.