解读：在经过人工验证的、相对标准的 Issue 修复任务上，Qwen3 并没有展现出统治力，反而是 MiniMax 这种黑马表现抢眼。这说明在“标准题”上，各家模型差异不大，甚至 Qwen3 还有点“偏科”。

An AI agent got nasty after its pull request got rejected. Can open-source development survive autonomous bot contributors?

智东西（公众号：zhidxcom）作者 | 李水青编辑 | 心缘智东西2月4日报道，今日凌晨，阿里开源了一款小型混合专家模型Qwen3-Coder-Next，专为编程智能体（Agent）和本地开发打造。该模型总参数80B，激活参数仅3B，在权威基准SWE-Bench ...

On SWE-Bench Verified, the model achieved a score of 70.6%. This performance is notably competitive when placed alongside significantly larger models; it outpaces DeepSeek-V3.2, which scores 70.2%, ...

Qwen Code’s Qwen3-Coder model doesn’t seem as good as its benchmark scores imply, but the tools are free and the usage limits are generous. The three biggest hyperscalers in the US are AWS, Microsoft ...

【TechWeb】7月23日消息，阿里通义千问发布迄今为止最具代理能力的代码模型Qwen3-Coder，并正式开源。 据介绍，Qwen3-Coder 拥有多个尺寸，当前最强大的版本Qwen3-Coder-480B-A35B-Instruct是一个 480B 参数激活 35B 参数的 MoE 模型，原生支持 256K token 的上下文并可通过 YaRN ...

Add Yahoo as a preferred source to see more of our stories on Google. Digital technology, software development concept. Coding programmer, software engineer working on laptop with circuit board and ...

AUSTIN, Texas--(BUSINESS WIRE)--Coder, the development platform that keeps developers in flow, today announced the general availability of Coder 2.0, an open source, cloud-native platform that allows ...