QW

Qwen3.7-Max-Preview

Reasoning modelQwen3.7

Qwen3.7-Max-Preview

Release date: 2026-05-20Updated: 2026-05-21 21:58:53.6063,810
Live demoGitHubHugging FaceCompare
Parameters
1T
Context length
1M
Chinese support
Supported
Reasoning ability

Qwen3.7-Max 是阿里云通义团队于2026年5月发布的闭源旗舰模型,定位为 Agent 工作流基座。模型在代码 Agent、通用 Agent 及长程自主执行方向进行了系统强化,在 GPQA Diamond(92.4)、HLE(41.4)、SWE-Pro(60.6)、MCP-Atlas(76.4)等主要基准上达到同批对比模型最高分,推理和 Agent 能力整体持平或小幅超越 Claude Opus 4.6 Max。官方实测显示模型可在未知硬件架构上持续自主运行 35 小时、执行逾千次工具调用,实现 10 倍算子加速。当前仅通过阿里云百炼平台 API 提供服务,兼容 OpenAI 与 Anthropic 两种调用协议。

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Qwen3.7-Max-Preview

Model basics

Reasoning traces
Supported
Thinking modes
Thinking Mode (Default)Standard Mode
Context length
1M tokens
Max output length
64K tokens
Model type
Reasoning model
Modality (in / out)
No data
Release date
2026-05-20
Model file size
No data
MoE architecture
Yes
Total params / Active params
1T / N/A
Knowledge cutoff
No data
Qwen3.7-Max-Preview

Open source & experience

Code license
不开源
Weights license
不开源
GitHub repo
GitHub link unavailable
Hugging Face
Hugging Face link unavailable
Qwen3.7-Max-Preview

Official resources

DataLearnerAI blog
No blog post yet
Qwen3.7-Max-Preview

API details

API speed
3/5
💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.
Standard
TypeConditionInputOutput
Text-$2.50/ 1M$7.50/ 1M
Cache PricingPrompt Cache
TypeTTLWriteRead
Text5m$3.13/ 1M$0.250/ 1M
Qwen3.7-Max-Preview

Benchmark Results

Qwen3.7-Max-Preview currently shows benchmark results led by MMLU Pro (4 / 126, score 89.60), LiveCodeBench (4 / 120, score 91.60), GPQA Diamond (11 / 179, score 92.40). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
Tool usage

General Knowledge

4 evaluations
Benchmark / mode
Score
Rank/total
92.40
11 / 179
89.60
4 / 126
HLE
Thinking ModeTools
53.50
12 / 161
HLE
Max
41.40
50 / 161

Coding and Software Engineer

4 evaluations
Benchmark / mode
Score
Rank/total
91.60
4 / 120
SWE-bench Verified
Thinking ModeTools
80.40
12 / 108
SWE-bench Multilingual
Thinking ModeTools
78.30
3 / 20
SWE-Bench Pro - Public
Thinking ModeTools
60.60
6 / 44

Instruction Following

1 evaluations
Benchmark / mode
Score
Rank/total
79.10
2 / 29

AI Agent - Tool Usage

1 evaluations
Benchmark / mode
Score
Rank/total
Terminal Bench 2.0
Thinking ModeTools
69.70
5 / 46

Math and Reasoning

1 evaluations
Benchmark / mode
Score
Rank/total
90
2 / 20

Compare with other models

Qwen3.7-Max-Preview

Publisher

Qwen3.7-Max-Preview

Model Overview

Qwen3.7-Max 是阿里云通义团队于2026年5月发布的闭源旗舰模型,定位为 Agent 工作流基座。模型在代码 Agent、通用 Agent 及长程自主执行方向进行了系统强化,在 GPQA Diamond(92.4)、HLE(41.4)、SWE-Pro(60.6)、MCP-Atlas(76.4)等主要基准上达到同批对比模型最高分,推理和 Agent 能力整体持平或小幅超越 Claude Opus 4.6 Max。官方实测显示模型可在未知硬件架构上持续自主运行 35 小时、执行逾千次工具调用,实现 10 倍算子加速。当前仅通过阿里云百炼平台 API 提供服务,兼容 OpenAI 与 Anthropic 两种调用协议。

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code