Qwen3.6-Max-Preview:更智能、更锐利,持续进化 | Qwen
Qwen3.6-Max-Preview:更智能、更锐利,持续进化
QWEN STUDIODISCORD
继 Qwen3.6-Plus 发布之后,我们现分享下一个自研模型的早期预览版:Qwen3.6-Max-Preview。与 Qwen3.6-Plus 相比,此预览版本带来了更强的世界知识和指令遵循能力,并在广泛的基准测试中实现了显著的智能体编程改进。作为预览版,该模型仍在积极开发中——我们正在持续迭代,并预计后续版本将取得进一步提升。
Qwen3.6-Max-Preview 是通过阿里云百炼平台提供的托管自研模型,其特点包括:相较于 Qwen3.6-Plus 更强的智能体编程能力、更丰富的世界知识和指令遵循能力,以及改进的真实世界智能体与知识可靠性表现。
您可以在 Qwen Studio 上进行交互式对话,或通过阿里云百炼 API 以 `qwen3.6-max-preview` 名称调用(即将推出)。
性能表现#
以下我们展示了 Qwen3.6-Max-Preview 与领先前沿模型的评估对比。与 Qwen3.6-Plus 相比,此预览版本在智能体编程方面取得了显著提升(例如 SkillsBench +9.9,SciCode +6.3,NL2Repo +5.0,Terminal-Bench 2.0 +3.8),更强的世界知识(SuperGPQA +2.3,QwenChineseBench +5.3),以及更好的指令遵循能力(ToolcallFormatIFBench +2.8)。
使用 Qwen3.6-Max-Preview 进行构建#
Qwen3.6-Max-Preview 即将登陆阿里云百炼平台。请耐心等待,直至我们完全准备就绪。
Qwen3.6-Max-Preview 可通过阿里云百炼 API 以 `qwen3.6-max-preview` 名称使用。您也可以在 Qwen Studio 上立即试用。
API 使用#
此版本支持 preserve_thinking 功能:保留消息中所有先前轮次的思考内容,建议在智能体任务中使用。
阿里云百炼#
阿里云百炼支持行业标准协议,包括兼容 OpenAI 规范的聊天补全和响应 API,以及兼容 Anthropic 的 API 接口。
以下是聊天补全 API 的示例代码:
python
环境变量(根据官方文档):DASHSCOPEAPIKEY:您的 API 密钥,来自 https://modelstudio.console.alibabacloud.com;DASHSCOPEBASEURL:(可选)兼容模式 API 的基础 URL。
- 北京:https://dashscope.aliyuncs.com/compatible-mode/v1
- 新加坡:https://dashscope-intl.aliyuncs.com/compatible-mode/v1
- 美国(弗吉尼亚):https://dashscope-us.aliyuncs.com/compatible-mode/v1;DASHSCOPEMODEL:(可选)模型名称;可覆盖为不同模型。"""from openai import OpenAIimport osapikey = os.environ.get("DASHSCOPEAPIKEY")if not apikey: raise ValueError( "DASHSCOPEAPIKEY is required. " "Set it via: export DASHSCOPEAPIKEY='your-api-key'" )client = OpenAI( apikey=apikey, baseurl=os.environ.get( "DASHSCOPEBASEURL", "https://dashscope-intl.aliyuncs.com/compatible-mode/v1", ),)messages = [{"role": "user", "content": "Introduce vibe coding."}]model = os.environ.get( "DASHSCOPEMODEL", "qwen3.6-max-preview",)completion = client.chat.completions.create( model=model, messages=messages, extrabody={ "enablethinking": True, # "preservethinking": True, }, stream=True)reasoningcontent = "" # 完整推理过程answercontent = "" # 完整回复isanswering = False # 是否已进入回答阶段print("\n" + "=" 20 + "Reasoning" + "=" 20 + "\n")for chunk in completion: if not chunk.choices: print("\nUsage:") print(chunk.usage) continue delta = chunk.choices[0].delta # 仅当存在推理内容时收集 if hasattr(delta, "reasoningcontent") and delta.reasoningcontent is not None: if not isanswering: print(delta.reasoningcontent, end="", flush=True) reasoningcontent += delta.reasoningcontent # 收到内容,开始回答阶段 if hasattr(delta, "content") and delta.content: if not isanswering: print("\n" + "=" 20 + "Answer" + "=" 20 + "\n") isanswering = True print(delta.content, end="", flush=True) answercontent += delta.content
更多信息,请访问 API 文档。
摘要
Qwen3.6-Max-Preview 是我们下一代专有模型的早期预览版,在智能体编程、世界知识和指令遵循方面相比 Qwen3.6-Plus 带来了显著改进。它在六大主要编程基准测试——SWE-bench Pro、Terminal-Bench 2.0、SkillsBench、QwenClawBench、QwenWebBench 和 SciCode——上均取得了最高分,相比前代模型有大幅提升。同时,它还展现了更强的知识能力(SuperGPQA、QwenChineseBench)和更优的指令遵循能力(ToolcallFormatIFBench)。
作为预览版本,Qwen3.6-Max-Preview 仍在积极开发中。我们将继续迭代该模型,并期待在后续版本中实现进一步改进。我们欢迎社区反馈,并期待看到大家的创作。敬请关注!
引用#
如果您认为 Qwen3.6-Max-Preview 有帮助,欢迎引用以下文章:
bibtex
@misc{qwen36maxpreview, title = {{Qwen3.6-Max-Preview}:更智能、更锐利,仍在进化}, url = {https://qwen.ai/blog?id=qwen3.6-max-preview}, author = {{Qwen 团队}}, month = {四月}, year = {2026}}
")
Android
macOS
Windows
Qwen Studio 概览
下载
Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving | Qwen
Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving
QWEN STUDIODISCORD
Following the release of Qwen3.6-Plus, we are sharing an early preview of our next proprietary model: Qwen3.6-Max-Preview. Compared to Qwen3.6-Plus, this preview release brings stronger world knowledge and instruction following, along with significant agentic coding improvements across a wide range of benchmarks. As a preview, the model is still under active development — we are continuing to iterate and expect further gains in subsequent versions.
Qwen3.6-Max-Preview is the hosted proprietary model available via Alibaba Cloud Model Studio, featuring: improved agentic coding capability over Qwen3.6-Plus stronger world knowledge and instruction following improved real-world agent and knowledge reliability performance
You can chat interactively on Qwen Studio or call via API as qwen3.6-max-preview on Alibaba Cloud Model Studio API (coming soon).
Performance#
Below we present evaluations of Qwen3.6-Max-Preview against leading frontier models. Compared to Qwen3.6-Plus, the preview release delivers significant improvements in agentic coding (e.g., SkillsBench +9.9, SciCode +6.3, NL2Repo +5.0, Terminal-Bench 2.0 +3.8), stronger world knowledge (SuperGPQA +2.3, QwenChineseBench +5.3), and better instruction following (ToolcallFormatIFBench +2.8).
Build with Qwen3.6-Max-Preview#
Qwen3.6-Max-Preview is coming soon to Alibaba Cloud Model Studio. Please stand by until we are fully ready.
Qwen3.6-Max-Preview is available through the Alibaba Cloud Model Studio API as qwen3.6-max-preview. You can also try it instantly on Qwen Studio.
API Usage#
This release supports the preservethinking feature: preserving thinking content from all preceding turns in messages, which is recommended for agentic tasks.
Alibaba Cloud Model Studio#
Alibaba Cloud Model Studio supports industry-standard protocols, including chat completions and responses APIs compatible with OpenAI’s specification, as well as an API interface compatible with Anthropic.
Example code for chat completions API is provided below:
python
""" Environment variables (per official docs): DASHSCOPEAPIKEY: Your API Key from https://modelstudio.console.alibabacloud.com DASHSCOPEBASEURL: (optional) Base URL for compatible-mode API.
- Beijing: https://dashscope.aliyuncs.com/compatible-mode/v1
- Singapore: https://dashscope-intl.aliyuncs.com/compatible-mode/v1
- US (Virginia): https://dashscope-us.aliyuncs.com/compatible-mode/v1 DASHSCOPEMODEL: (optional) Model name; override for different models. """from openai import OpenAIimport osapikey = os.environ.get("DASHSCOPEAPIKEY")if not apikey: raise ValueError( "DASHSCOPEAPIKEY is required. " "Set it via: export DASHSCOPEAPIKEY='your-api-key'" )client = OpenAI( apikey=apikey, baseurl=os.environ.get( "DASHSCOPEBASEURL", "https://dashscope-intl.aliyuncs.com/compatible-mode/v1", ),)messages = [{"role": "user", "content": "Introduce vibe coding."}]model = os.environ.get( "DASHSCOPEMODEL", "qwen3.6-max-preview",)completion = client.chat.completions.create( model=model, messages=messages, extrabody={ "enablethinking": True, # "preservethinking": True, }, stream=True)reasoningcontent = "" # Full reasoning traceanswercontent = "" # Full responseisanswering = False # Whether we have entered the answer phaseprint("\n" + "=" 20 + "Reasoning" + "=" 20 + "\n")for chunk in completion: if not chunk.choices: print("\nUsage:") print(chunk.usage) continue delta = chunk.choices[0].delta # Collect reasoning content only if hasattr(delta, "reasoningcontent") and delta.reasoningcontent is not None: if not isanswering: print(delta.reasoningcontent, end="", flush=True) reasoningcontent += delta.reasoningcontent # Received content, start answer phase if hasattr(delta, "content") and delta.content: if not isanswering: print("\n" + "=" 20 + "Answer" + "=" 20 + "\n") isanswering = True print(delta.content, end="", flush=True) answercontent += delta.content
For more information, please visit the API doc.
Summary#
Qwen3.6-Max-Preview is an early preview of our next proprietary model, delivering meaningful improvements over Qwen3.6-Plus in agentic coding, world knowledge, and instruction following. It achieves the top score on six major coding benchmarks — SWE-bench Pro, Terminal-Bench 2.0, SkillsBench, QwenClawBench, QwenWebBench, and SciCode — with substantial gains over its predecessor. It also demonstrates stronger knowledge (SuperGPQA, QwenChineseBench) and better instruction following (ToolcallFormatIFBench).
As a preview release, Qwen3.6-Max-Preview is still under active development. We are continuing to iterate on the model and expect further improvements in subsequent versions. We welcome community feedback and look forward to seeing what you build. Stay tuned!
Citation#
Feel free to cite the following article if you find Qwen3.6-Max-Preview helpful:
bibtex
@misc{qwen36maxpreview, title = {{Qwen3.6-Max-Preview}: Smarter, Sharper, Still Evolving}, url = {https://qwen.ai/blog?id=qwen3.6-max-preview}, author = {{Qwen Team}}, month = {April}, year = {2026}}
")
Android
macOS
Windows
Qwen Studio Overview
Download