阿里巴巴云千问团队发布Qwen3.7-Max，这是一款专为长时间自主智能体任务设计的闭源模型。该模型曾在单次会话中持续运行35小时，为阿里巴巴自研的专用芯片优化代码。在标准基准测试中，Qwen3.7-Max的表现与Claude Opus 4.6持平，并超越了国内竞品如DeepSeek V4 Pro和Kimi K2.6。此外，团队还现场演示了该模型指挥一个四足机器人执行任务的能力。
原文 · 未翻译
Alibaba's latest AI model ran autonomously for 35 hours to optimize code for its own custom chip
Alibaba's Qwen team has released Qwen3.7-Max, a proprietary model designed for agent-based tasks. In a real-world test, the model ran a fully autonomous kernel optimization for 35 hours straight.
Like its predecessors Qwen3-Max and Qwen3.6-Plus, the new Max version is only available through the Alibaba Cloud Model Studio API. Alibaba used to release its Qwen models as open source, but that's changed. The last open flagship was Qwen3.5-397B-A17B from February 2026.
Qwen3.7-Max supports OpenAI- and Anthropic-compatible interfaces and plugs right into Claude Code, OpenClaw, or Qwen Code. The Qwen team says the model targets four use cases: working as a coding agent from front-end prototypes to complex multi-file software projects, automating office tasks with external tools, running autonomously for long stretches, and performing consistently across different agent frameworks.
A kernel experiment that ran for 35 hours
Qwen3.7-Max was tasked with optimizing a hardware-based attention kernel for the open-source inference software SGLang. The hardware was a cloud instance with T-Head-ZW-M890 accelerators, an AI chip platform from Alibaba's own semiconductor arm.
The Qwen team says the model had never seen this chip architecture during training. It started with no measurement data, no hardware docs, and no sample code. The only thing it had to work with was the existing reference implementation, written in the Triton programming language.
Over about 35 hours of nonstop autonomous work, the model ran 432 kernel tests with 1,158 total tool calls. It compiled, measured, and revised the code in loops, caught compilation errors, and tracked down performance bottlenecks on its own. The result, according to the Qwen researchers, is an average 10x speedup over the reference implementation.
Competitor models came up well short in the same setup. GLM 5.1 hit a 7.3x speedup, Kimi K2.6 got to 5x, DeepSeek V4 Pro managed 3.3x, and the predecessor Qwen3.6-Plus barely moved the needle at 1.1x. Models that quit early ended their sessions on their own after five straight rounds with no tool calls. On the standardized KernelBench L3 benchmark, Qwen3.7-Max claims to produce accelerated kernels 96 percent of the time, just behind Anthropic's Opus 4.6 at 98 percent.
阿里巴巴云千问团队发布Qwen3.7-Max，这是一款专为长时间自主智能体任务设计的闭源模型。该模型曾在单次会话中持续运行35小时，为阿里巴巴自研的专用芯片优化代码。在标准基准测试中，Qwen3.7-Max的表现与Claude Opus 4.6持平，并超越了国内竞品如DeepSeek V4 Pro和Kimi K2.6。此外，团队还现场演示了该模型指挥一个四足机器人执行任务的能力。
原文 · 保持原样，未翻译
Alibaba's latest AI model ran autonomously for 35 hours to optimize code for its own custom chip
Alibaba's Qwen team has released Qwen3.7-Max, a proprietary model designed for agent-based tasks. In a real-world test, the model ran a fully autonomous kernel optimization for 35 hours straight.
Like its predecessors Qwen3-Max and Qwen3.6-Plus, the new Max version is only available through the Alibaba Cloud Model Studio API. Alibaba used to release its Qwen models as open source, but that's changed. The last open flagship was Qwen3.5-397B-A17B from February 2026.
Qwen3.7-Max supports OpenAI- and Anthropic-compatible interfaces and plugs right into Claude Code, OpenClaw, or Qwen Code. The Qwen team says the model targets four use cases: working as a coding agent from front-end prototypes to complex multi-file software projects, automating office tasks with external tools, running autonomously for long stretches, and performing consistently across different agent frameworks.
Training splits task, tool environment, and validator
Qwen3.7-Max builds on a training approach the team first rolled out with Qwen3.5. Each training task breaks into three independent pieces: the actual task, the tool environment, and the validator that checks the result. These can be mixed and matched freely.
The same task gets practiced across different tool environments and checked with different test methods. That's meant to force the model to pick up strategies that work everywhere, not just shortcuts tied to one specific setup. On QwenClawBench and CoWorkBench, Qwen3.7-Max holds steady no matter which test environment it's dropped into, the team says.
The model polices its own training for reward hacking
The Qwen team also put Qwen3.7-Max to work as a watchdog during its own training. The model watched training runs for software engineering tasks for over 80 hours and ran more than 10,000 checks. It hunted for tricks the model being trained might pull to game its rewards, like grabbing correct answers straight off GitHub. Qwen3.7-Max wrote 13 new detection rules and flagged 1,618 cases.
One year in simulation tests long-term planning
To gauge long-term planning, the team used YC-Bench, a benchmark that simulates a startup's full one-year life cycle. The model has to manage staff across hundreds of decision rounds, review contracts, spot bad-faith customers, and keep profit margins healthy against rising labor costs.
Qwen3.7-Max pulled in $2.08 million in total revenue and wrapped up 237 tasks. Its predecessor, Qwen3.6-Plus, hit $1.05 million. Qwen3.5-Plus managed just $352,000.
Across most benchmarks, Qwen3.7-Max trades blows with Claude Opus 4.6 Max, Kimi K2.6 Thinking, GLM-5.1 Thinking, and DeepSeek V4 Pro Max. On SWE-Verified, the model scored 80.4, nearly tied with Opus 4.6 Max (80.8) and DeepSeek V4 Pro Max (80.6). On the math and science benchmarks GPQA Diamond (92.4), HMMT 2026 February (97.1), and Apex (44.5), Qwen3.7-Max tops the provider's own comparison table.
Some of those benchmarks are homegrown, though. QwenWebDev, QwenClawBench, CoWorkBench, and QwenWorldBench all come from the Qwen team itself. Every result here is self-reported. A closer look at scaling dynamics and methodology is coming in an upcoming technical report.
Beyond the usual use cases, the team also shows off Qwen3.7-Max steering a four-legged robot. Using its own robotics framework and a paired navigation model, the language model guides the robot through physical spaces.
Qwen3.7-Max was tasked with optimizing a hardware-based attention kernel for the open-source inference software SGLang. The hardware was a cloud instance with T-Head-ZW-M890 accelerators, an AI chip platform from Alibaba's own semiconductor arm.
The Qwen team says the model had never seen this chip architecture during training. It started with no measurement data, no hardware docs, and no sample code. The only thing it had to work with was the existing reference implementation, written in the Triton programming language.
Over about 35 hours of nonstop autonomous work, the model ran 432 kernel tests with 1,158 total tool calls. It compiled, measured, and revised the code in loops, caught compilation errors, and tracked down performance bottlenecks on its own. The result, according to the Qwen researchers, is an average 10x speedup over the reference implementation.
Competitor models came up well short in the same setup. GLM 5.1 hit a 7.3x speedup, Kimi K2.6 got to 5x, DeepSeek V4 Pro managed 3.3x, and the predecessor Qwen3.6-Plus barely moved the needle at 1.1x. Models that quit early ended their sessions on their own after five straight rounds with no tool calls. On the standardized KernelBench L3 benchmark, Qwen3.7-Max claims to produce accelerated kernels 96 percent of the time, just behind Anthropic's Opus 4.6 at 98 percent.
Training splits task, tool environment, and validator
Qwen3.7-Max builds on a training approach the team first rolled out with Qwen3.5. Each training task breaks into three independent pieces: the actual task, the tool environment, and the validator that checks the result. These can be mixed and matched freely.
The same task gets practiced across different tool environments and checked with different test methods. That's meant to force the model to pick up strategies that work everywhere, not just shortcuts tied to one specific setup. On QwenClawBench and CoWorkBench, Qwen3.7-Max holds steady no matter which test environment it's dropped into, the team says.
The model polices its own training for reward hacking
The Qwen team also put Qwen3.7-Max to work as a watchdog during its own training. The model watched training runs for software engineering tasks for over 80 hours and ran more than 10,000 checks. It hunted for tricks the model being trained might pull to game its rewards, like grabbing correct answers straight off GitHub. Qwen3.7-Max wrote 13 new detection rules and flagged 1,618 cases.
One year in simulation tests long-term planning
To gauge long-term planning, the team used YC-Bench, a benchmark that simulates a startup's full one-year life cycle. The model has to manage staff across hundreds of decision rounds, review contracts, spot bad-faith customers, and keep profit margins healthy against rising labor costs.
Qwen3.7-Max pulled in $2.08 million in total revenue and wrapped up 237 tasks. Its predecessor, Qwen3.6-Plus, hit $1.05 million. Qwen3.5-Plus managed just $352,000.
Across most benchmarks, Qwen3.7-Max trades blows with Claude Opus 4.6 Max, Kimi K2.6 Thinking, GLM-5.1 Thinking, and DeepSeek V4 Pro Max. On SWE-Verified, the model scored 80.4, nearly tied with Opus 4.6 Max (80.8) and DeepSeek V4 Pro Max (80.6). On the math and science benchmarks GPQA Diamond (92.4), HMMT 2026 February (97.1), and Apex (44.5), Qwen3.7-Max tops the provider's own comparison table.
Some of those benchmarks are homegrown, though. QwenWebDev, QwenClawBench, CoWorkBench, and QwenWorldBench all come from the Qwen team itself. Every result here is self-reported. A closer look at scaling dynamics and methodology is coming in an upcoming technical report.
Beyond the usual use cases, the team also shows off Qwen3.7-Max steering a four-legged robot. Using its own robotics framework and a paired navigation model, the language model guides the robot through physical spaces.