llm 0.32a2

The llm team has released version 0.32a2, featuring significant enhancements for OpenAI models. Most reasoning-capable models now utilize the /v1/responses endpoint instead of /v1/chat/completions, facilitating interleaved reasoning across tool calls. This update allows users to view summarized reasoning tokens when executing prompts against OpenAI models, providing insight into how the model processes inputs. Users can opt to hide these tokens using the -R or --hide-reasoning flags if desired.