llm 0.32rc1

AI 中文解读

AI工具LLM发布新版本，最亮眼的是给对话记录装上了“记忆整理术”——以后AI聊天的内容能像文件一样分类存放，重复内容自动合并，还能支持分叉对话，就像给聊天画了棵“关系树”，理清各种话题分支。普通用户没太大感知，但用AI开发应用的人会明显感到后台数据更清爽，查找历史记录更快。新版还特别提醒用户升级前先备份聊天日志，步骤很简单，一条命令就能搞定。另外，这次更新赶上了新模型热潮，支持了GPT-5.6系列的三款新模型，意味着用AI写作、编程时会获得更聪明的答案。对于天天用AI工具打工人和开发者来说，稳定性和效率提升会是实实在在的体验升级。

Release: <a href="https://github.com/simonw/llm/releases/tag/0.32rc1">llm 0.32rc1</a> This RC for LLM 0.32 finishes the work that <a href="https://simonwillison.net/2026/Apr/29/llm/">started in LLM 0.32a0</a> - it adds a <a href="https://llm.datasette.io/en/latest/logging.html#the-message-store">new schema design</a> that does a much better job of capturing the details of the prompts and responses returned by the latest model families. The most important change is the use of content-addressable hash IDs for stored messages. This allows de-duplication in the database, and means that LLM can now represent trees of messages for forked conversations. Since it involves a significant schema change - new tables only, and old data should not be affected at all - it's worth running a backup of your existing <code>logs.db</code> before upgrading to the RC: <pre><code>llm logs backup logs-backup.db </code></pre> The RC also adds support for <code>gpt-5.6-sol</code>, <code>gpt-5.6-terra</code>, and <code>gpt-5.6-luna</code>. Tags: <a href="https://simonwillison.net/tags/llm">llm</a>