AI Daily · 2026-08-23

Simon Willison released llm 0.33, focused on dependency modernization and more flexible configuration. The update moves to the OpenAI Python library 3.x and switches the HTTP client from httpx to httpx2, while the embed and embed-multi commands now accept per-call keys for embedding plugins. Templates can be combined through repeated -t/--template flags, making it easier to package model configurations and default options. For Responses API models with reasoning support, a new reasoning_summary option offers auto, concise, or detailed control. Overall, this release reduces maintenance overhead and improves embedding, templating, and reasoning workflows.

Ecosystem & Beyond (Products / Agents / Tools / Opinions)

Product Update

⭐⭐ [Product Update] llm 0.33

Simon Willison released llm 0.33, a command-line tool for accessing large language models. This release upgrades to the OpenAI Python library 3.x and switches the HTTP client dependency from httpx to httpx2. The llm embed and llm embed-multi commands and their Python methods now accept --key/key to pass per-call keys to embedding plugins. The -t/--template flag can be repeated to combine templates, enabling patterns that package a model with default options. Additionally, reasoning-capable Responses API models now support a reasoning_summary option with auto, concise, and detailed values.