llm-openai-decisions 0.1a0

Simon Willison released llm-openai-decisions 0.1a0, a plugin for the LLM tool that talks to OpenAI's new Decisions API. According to the release note, OpenAI announced the API previously at last week's DevDay.

Willison writes that he already maintains an llm-typesafe plugin for talking to Jev, and that he had GPT-6 Astra read the new OpenAI API documentation and build the plugin inspired by llm-typesafe.

The note states that, unlike Jev, the new gpt-6-luna decision model supports image input in addition to text. Both models charge for input tokens and not for output: OpenAI's is 10 cents per million input tokens, while Jev's is 4.2 cents per million. Willison describes the API shape as otherwise very similar to Jev, at least conceptually, and says Jev supports three question types, for yes/no, choices, or scores, with OpenAI Decisions supporting the same three types.

The plugin installs with the command "llm install llm-openai-decisions". The release note includes an example query against an image attachment using the model openai-decisions/gpt-6-luna and the prompt "Does this image contain any mammals?", with example output of a predicate named evaluation and a probability of 0.0. Willison directs readers to the README for full details on running the other types of questions. The release is tagged openai, llm, coding-agents and jev.