Poolside AI推出了两款新的开源权重代理编码模型Laguna XS.2和M.1。这些模型在SWE-bench Verified基准测试中取得了令人印象深刻的分数，M.1达到72.5%，XS.2达到68.2%。XS.2模型是一个拥有330亿参数的模型，其能够在拥有36GB内存的消费级硬件上本地运行，这使得先进的编码助手更加易于获取。 AI
影响 提供易于获取、可在本地运行的代理编码工具，可能降低开发者的门槛。
排序理由 发布具有基准测试结果的开源权重编码模型。
在 Mastodon — fosstodon.org 阅读 →
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
Poolside AI has released Laguna XS.2 and M.1, open-weight agentic coding models that achieve 68.2% and 72.5% on SWE-bench Verified respectively. The XS.2 model runs locally on a Mac with 36GB RAM. https://www. marktechpost.com/2026/04/28/po olside-ai-introduces-laguna-xs-2-and-m-…
American AI startup Poolside has released Laguna XS.2, a free open-source model for local agentic coding. The 33bn-parameter model runs locally on a single GPU, achieving 44.5% on SWE-bench Pro, nearly matching models six times its size. Available via Hugging Face and Ollama. htt…