<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？]]></title><description><![CDATA[<p dir="auto">最近想给团队接一个代码助手，不是聊天问答，是能读 repo、改文件、跑测试那种。DeepSeek、Qwen、Claude、GPT 怎么选？</p>
]]></description><link>https://localaihub.com/topic/79/代码助手场景-deepseek-qwen-claude-gpt-谁更像同事</link><generator>RSS for Node</generator><lastBuildDate>Wed, 03 Jun 2026 18:50:43 GMT</lastBuildDate><atom:link href="https://localaihub.com/topic/79.rss" rel="self" type="application/rss+xml"/><pubDate>Mon, 04 May 2026 14:42:00 GMT</pubDate><ttl>60</ttl><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Tue, 05 May 2026 08:10:00 GMT]]></title><description><![CDATA[<p dir="auto">对，别只统计“答案像不像”。代码助手最重要指标是 merge 后有没有新债。</p>
]]></description><link>https://localaihub.com/post/468</link><guid isPermaLink="true">https://localaihub.com/post/468</guid><dc:creator><![CDATA[小陈在改bug]]></dc:creator><pubDate>Tue, 05 May 2026 08:10:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Tue, 05 May 2026 06:18:00 GMT]]></title><description><![CDATA[<p dir="auto">那我先做 20 个真实 issue 回放，每个模型必须读文件、改 patch、跑测试。用成功率和人工返工分钟数排。</p>
]]></description><link>https://localaihub.com/post/467</link><guid isPermaLink="true">https://localaihub.com/post/467</guid><dc:creator><![CDATA[阿树]]></dc:creator><pubDate>Tue, 05 May 2026 06:18:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Tue, 05 May 2026 06:00:00 GMT]]></title><description><![CDATA[<p dir="auto">Claude/GPT 贵，但是能减少返工；Qwen/DeepSeek 便宜，但是要加 harness。不是谁替代谁，是组合方式。</p>
]]></description><link>https://localaihub.com/post/466</link><guid isPermaLink="true">https://localaihub.com/post/466</guid><dc:creator><![CDATA[momo]]></dc:creator><pubDate>Tue, 05 May 2026 06:00:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Tue, 05 May 2026 03:59:00 GMT]]></title><description><![CDATA[<p dir="auto">我踩过一个坑：长上下文塞整个 repo，模型越看越飘。后来改成检索相关文件 + 明确 diff 范围，准确率上去了。</p>
]]></description><link>https://localaihub.com/post/465</link><guid isPermaLink="true">https://localaihub.com/post/465</guid><dc:creator><![CDATA[小李不困]]></dc:creator><pubDate>Tue, 05 May 2026 03:59:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Tue, 05 May 2026 03:28:00 GMT]]></title><description><![CDATA[<p dir="auto">评测别用 HumanEval 单项。真实场景要有“读旧代码、按风格修改、跑失败测试、再修”的链路。</p>
]]></description><link>https://localaihub.com/post/464</link><guid isPermaLink="true">https://localaihub.com/post/464</guid><dc:creator><![CDATA[oneMoreTry]]></dc:creator><pubDate>Tue, 05 May 2026 03:28:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Tue, 05 May 2026 01:33:00 GMT]]></title><description><![CDATA[<p dir="auto">我们内部是小任务 Qwen，本地隐私代码也用它；跨模块重构用 Claude/GPT。DeepSeek 放在方案推理和 code review 前置。</p>
]]></description><link>https://localaihub.com/post/463</link><guid isPermaLink="true">https://localaihub.com/post/463</guid><dc:creator><![CDATA[青菜]]></dc:creator><pubDate>Tue, 05 May 2026 01:33:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Mon, 04 May 2026 23:32:00 GMT]]></title><description><![CDATA[<p dir="auto">不能全靠提示。要让执行器只允许它报告真实命令结果，最后产物里引用测试输出。模型自己说的不算。</p>
]]></description><link>https://localaihub.com/post/462</link><guid isPermaLink="true">https://localaihub.com/post/462</guid><dc:creator><![CDATA[林小北]]></dc:creator><pubDate>Mon, 04 May 2026 23:32:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Mon, 04 May 2026 22:01:00 GMT]]></title><description><![CDATA[<p dir="auto">这个能靠系统提示解决吗？</p>
]]></description><link>https://localaihub.com/post/461</link><guid isPermaLink="true">https://localaihub.com/post/461</guid><dc:creator><![CDATA[普通网友A]]></dc:creator><pubDate>Mon, 04 May 2026 22:01:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Mon, 04 May 2026 20:25:00 GMT]]></title><description><![CDATA[<p dir="auto">“谁更像同事”还有一个维度：会不会承认没跑测试。我们之前一个模型没跑也写“测试通过”，直接拉黑。</p>
]]></description><link>https://localaihub.com/post/460</link><guid isPermaLink="true">https://localaihub.com/post/460</guid><dc:creator><![CDATA[rootless]]></dc:creator><pubDate>Mon, 04 May 2026 20:25:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Mon, 04 May 2026 20:17:00 GMT]]></title><description><![CDATA[<p dir="auto">GPT 系列现在做 code agent 也成熟，尤其和工具链配合。成本会高，但少改坏一次生产代码就回来了。</p>
]]></description><link>https://localaihub.com/post/459</link><guid isPermaLink="true">https://localaihub.com/post/459</guid><dc:creator><![CDATA[陈一]]></dc:creator><pubDate>Mon, 04 May 2026 20:17:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Mon, 04 May 2026 20:03:00 GMT]]></title><description><![CDATA[<p dir="auto">Qwen Coder 系列我用在本地代码补全不错，中文注释也自然。真实 agent 还是要看工具调用和上下文管理。</p>
]]></description><link>https://localaihub.com/post/458</link><guid isPermaLink="true">https://localaihub.com/post/458</guid><dc:creator><![CDATA[小高]]></dc:creator><pubDate>Mon, 04 May 2026 20:03:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Mon, 04 May 2026 19:37:00 GMT]]></title><description><![CDATA[<p dir="auto">DeepSeek-R1 推理解释多，排查复杂 bug 有帮助。但接工具时要看响应格式稳定性，别让思考文本污染 patch 或 JSON。</p>
]]></description><link>https://localaihub.com/post/457</link><guid isPermaLink="true">https://localaihub.com/post/457</guid><dc:creator><![CDATA[mxm]]></dc:creator><pubDate>Mon, 04 May 2026 19:37:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Mon, 04 May 2026 17:18:00 GMT]]></title><description><![CDATA[<p dir="auto">Claude 在长文件阅读和改动解释上挺强，尤其大 repo。我不喜欢的是有时太愿意写大方案，得把指令压紧。</p>
]]></description><link>https://localaihub.com/post/456</link><guid isPermaLink="true">https://localaihub.com/post/456</guid><dc:creator><![CDATA[JasonL]]></dc:creator><pubDate>Mon, 04 May 2026 17:18:00 GMT</pubDate></item><item><title><![CDATA[Reply to 代码助手场景，DeepSeek、Qwen、Claude、GPT 谁更像同事？ on Mon, 04 May 2026 17:00:00 GMT]]></title><description><![CDATA[<p dir="auto">能不能像同事，关键不是只会写代码，是会不会先读现有结构。很多模型写法漂亮，但不尊重项目边界。</p>
]]></description><link>https://localaihub.com/post/455</link><guid isPermaLink="true">https://localaihub.com/post/455</guid><dc:creator><![CDATA[小陈在改bug]]></dc:creator><pubDate>Mon, 04 May 2026 17:00:00 GMT</pubDate></item></channel></rss>