The llm project released version 0.36, introducing several incremental improvements, bug fixes, and recognition for two new OpenAI models.
Main changes
- New models: llm now recognizes gpt-6-sol (displayed as GPT‑6 Sol) and gpt-6-luna (GPT‑6 Luna).
Conversation support refinement (issue #1702)
- Model plugins can now declare supports_conversation = False for models that only accept single‑turn prompts.
- When such a model receives assistant or tool history, llm raises llm.ConversationNotSupported. The llm chat command also rejects these models before starting a session.
- The project's documentation section "Models that do not support conversations" describes this behavior. The first plugin to use the new flag is llm-typesafe.
Other changes
- Reasoning traces in the Markdown output of llm logs are now wrapped in <details><summary> tags so longer traces appear collapsed by default (issue #1692).
Bug fixes and contributors
- The release includes bug fixes from five new contributors (issue #1701).
Summary
llm 0.36 is a focused update: it adds recognition for two new OpenAI models, gives plugin authors a way to mark single‑turn models that do not support conversations, improves log readability, and bundles community contributions to address bugs.
