AI BriefingLLM 0.32a2 Released
Key point
LLM 0.32a2 adds support for the OpenAI Responses API and displays reasoning tokens.
Details
LLM, a tool for working with large language models from the command line, has released alpha version 0.32a2. The key change is that OpenAI's reasoning models now use the /v1/responses endpoint.
Using the Responses API instead of the previous /v1/chat/completions enables interleaved reasoning, where models like GPT-5 can continue reasoning between tool calls. This is the core change in this release.
Display of reasoning tokens has also been added.
- When you run a prompt with an OpenAI model, a summary of reasoning tokens is displayed in a different color on standard error (stderr).
- If you don't want this, you can hide it with
-Ror--hide-reasoning.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.