精选理由
Simon Willison 分享了 Kimi K3 的实测感受，还聊了 pelican 基准的局限性，想了解模型真实水平可以看看。
Simon Willison 发布关于 Kimi K3 模型的个人笔记，指出尽管 pelican 基准测试与模型实际能力（如长对话中的智能体工具调用）脱节，但仍能从中获得洞察。pelican 基准当前得分已无法完全反映 Kimi K3 在复杂任务上的表现。Willison 强调模型在 agentic tool calling 和长上下文交互上的进步比基准分数更重要。
原文 · Simon Willison
My notes on Kimi K3, plus some thoughts on what we can still learn from the pelican benchmark even w...
My notes on Kimi K3, plus some thoughts on what we can still learn from the pelican benchmark even while it becomes further detached from how good the models are at the things that matter (like agentic tool calling across longer conversations) simonwillison.net/2026/Jul/16/ki… 💬 12 🔄 8 ❤️ 85 👀 4894 📊 23 ⚡