LLM 0.32.1 Fixes Broken Installs Caused by OpenAI Update

Developer Simon Willison has released version 0.32.1 of his command-line utility, LLM, to address a critical bug that recently broke fresh installations of the tool. The issue arose after the official OpenAI Python library updated and dropped its internal usage of the httpx library. Because the LLM tool relied on httpx but only installed it transitively through its dependency on the OpenAI library, the sudden removal of httpx by OpenAI left new installations of LLM unable to function.

To resolve the issue immediately, the new 0.32.1 dot-release implements a temporary workaround by pinning the required OpenAI dependency to version "openai<3". This ensures that fresh installations of LLM will pull in an older, compatible version of the OpenAI library that still includes the necessary transitive dependencies. Willison also announced that a subsequent 0.33 release is planned to drop soon, which will permanently transition the tool's underlying HTTP client architecture from httpx to httpx2.

For developers and command-line practitioners who rely on the LLM tool to interact with large language models directly from their terminals, this update restores immediate installation stability. Users who experienced failed setups over the past few days should update to version 0.32.1 to restore functionality. The upcoming transition to httpx2 in version 0.33 highlights the ongoing maintenance challenges of managing transitive dependencies in rapidly evolving AI software ecosystems.