project
Qwen3.6-Max-Preview - The next-generation flagship model launched by Ali Tongyi
Qwen3.6-Max-Preview is an early preview version of the next-generation flagship large-scale model released by Alibaba's Tongyi Qianwen team. Compared to Qwen3.6-Plus, the model achieves significant improvements in three major dimensions: agent programming, world knowledge, and instruction compliance...
What is Qwen3.6-Max-Preview?
Qwen3.6-Max-Preview is an early preview version of the next-generation flagship model released by the Alibaba Tongyi Qianwen team. Compared to Qwen3.6-Plus, the model has achieved significant improvements in three dimensions: agent programming, world knowledge, and instruction compliance, achieving top scores in six programming benchmark tests including SWE-bench Pro, Terminal-Bench 2.0, and SciCode. Users can experience it online through Qwen Studio or by calling it through the Alibaba Cloud Bailian API.
Main functions of Qwen3.6-Max-Preview
-
Intelligent agent programmingIt achieved the highest score in six programming benchmarks, including SWE-bench Pro, Terminal-Bench 2.0, and SkillsBench, and supports code generation, terminal operation, and long-term code tasks.
-
World Knowledge EnhancementGraduate-level knowledge quiz (SuperGPQA) and Chinese knowledge (QwenChineseBench) skills have significantly improved.
-
Instruction compliance optimizationIn real-world scenarios, following the `ToolcallFormatIFBench` format yields better performance.
-
Mind chain retention:support
preserve_thinkingThis feature allows for the retention of prior thought processes during multi-turn dialogues, making it suitable for complex intelligent agent tasks. -
Protocol compatibilityThe API is compatible with OpenAI (chat completions/responses) and Anthropic interface specifications, facilitating seamless integration into existing workflows.
How to use Qwen3.6-Max-Preview
- Online experienceVisit Qwen Studio and select the Qwen3.6-Max-Preview model to begin the interactive dialogue.
- API calls
- Visit the Alibaba Cloud Bailian official website to obtain the API Key.
- When calling, fill in the model name.
qwen3.6-max-preview. - It supports OpenAI standard protocols (chat completions/responses) and the Anthropic API interface, allowing direct replacement of model names in existing workflows.
Key information and usage requirements of Qwen3.6-Max-Preview
-
Product Positioning: An early preview version of the next-generation flagship model of Ali Tongyi Qianwen, an upgraded version following Qwen3.6-Plus.
-
Core EnhancementThe intelligent agent's programming ability has been significantly enhanced, and its performance in following world knowledge and instructions is better.
-
PerformanceIt achieved the highest score in six programming benchmarks: SWE-bench Pro, Terminal-Bench 2.0, SkillsBench, QwenClawBench, QwenWebBench, and SciCode.
-
Status DescriptionIt is currently in preview and the model is being actively iterated. Subsequent versions will continue to optimize it.
The core advantages of Qwen3.6-Max-Preview
-
Leading in intelligent agent programmingIt achieved the highest scores in six programming benchmarks, including SWE-bench Pro, Terminal-Bench 2.0, and SkillsBench, and its code generation and terminal operation capabilities significantly surpassed its predecessor.
-
World Knowledge EnhancementThe performance of graduate-level knowledge Q&A (SuperGPQA) and Chinese knowledge (QwenChineseBench) has been significantly improved, and the reliability of the knowledge is better.
-
Instructions are followed more preciselyThe tool call format in real-world scenarios is more robust and adheres more strictly to the (ToolcallFormatIFBench) standard, resulting in more stable execution of complex tasks.
-
Competitors performed wellIt surpasses mainstream cutting-edge models such as Claude 4.5 Opus and GLM 5.1 on multiple benchmarks.
-
Strong ecological compatibilityThe API is compatible with the OpenAI and Anthropic protocols and supports...
preserve_thinkingThis feature allows for seamless integration into existing agent workflows.
Comparison of Qwen3.6-Max-Preview with similar competing products
| Comparison Dimensions | Qwen3.6-Max-Preview | Claude 4.5 Opus | GLM 5.1 |
|---|---|---|---|
| SWE-bench Pro(Project Code) | 57.3 | 50.9 | 58.4 |
| Terminal-Bench 2.0(Terminal programming) | 65.4 | 59.3 | 63.5 |
| SkillsBench(Agent skills) | 55.6 | 45.3 | 53.1 |
| SciCode(Research Code) | 47.0 | 49.5 | — |
| NL2Repo(Long-range code) | 42.9 | 43.2 | 42.7 |
| QwenWebBench(Front-end development) | 1532 | — | 1558 |
| SuperGPQA(Postgraduate knowledge) | 73.9 | 70.6 | — |
| QwenChineseBench(Chinese knowledge) | 84.0 | 69.0 | — |
| ToolcallFormatIFBench(Instructions followed) | 86.1 | 84.2 | 60.1 |
| GDPval-AA(Real-world mission) | 51.0 | 48.0 | 52.0 |
Application scenarios of Qwen3.6-Max-Preview
-
Software developmentSupports code generation, debugging, terminal operation, and repository-level long-term programming tasks, adapting to the full-stack development process.
-
Scientific computingIt assists in research-oriented code writing and data analysis, improving the efficiency of scientific research automation.
-
Front-end buildingGenerate web designs, web applications, games, SVGs, data visualizations, animations, and 3D content.
-
Intelligent agent automationThe model can perform complex real-world tasks and connects multiple workflows through tool calls.
-
Knowledge Q&AThe model provides graduate-level professional knowledge answers and in-depth knowledge services in the Chinese language field.