Summary & Key Takeaways ​

llm 0.36 adds support for OpenAI's new gpt-6-sol and gpt-6-luna models. Model plugins can now declare supports_conversation = False for single-turn models. The llm chat interface will reject models that do not support conversations. Reasoning traces in llm logs Markdown output are now wrapped in <details> tags.

Our Commentary ​

It's good to see llm keeping pace with the rapid model releases. The supports_conversation flag is a smart addition; it helps manage expectations for models not designed for multi-turn interactions. We appreciate the continuous refinement of developer tools in this fast-moving AI landscape.