OpenAI opens its Decisions API to all developers in public beta, charging only for input tokens

OpenAI made its Decisions API available to all developers in public beta on 6 October, in a post on its developer forum. Its documentation says: “The Decisions API is in public beta, and we expect to GA in the coming weeks.”

The endpoint is built for the small judgements an app or an agent makes many times over, such as which model, tool or action to use next. It takes text or image input and returns one of three typed outputs: predicates, which give the probability that a statement is true; choices, picked from a fixed list with a confidence; and scores on a numeric range.

gpt-6-luna is the only model available. Input costs $0.10 per million tokens, and the documentation says there are “no cache-read, cache-write, or output-token charges.”

OpenAI's speed claim is worded differently in its two places. The documentation says the API “returns typed answers about 10x faster than the Responses API”; the forum post says “up to 10x faster than GPT-6 Luna through the Responses API”. Nobody has published an independent measurement.

For developers building agents, the pricing matters as much as the speed: a routing or classification call that pays nothing for output can cost a fraction of a full chat completion.

• Confirmed Public beta for all developers from 6 October; general availability is expected in the coming weeks. OpenAI Developer Community
• Confirmed gpt-6-luna is the only model; input is $0.10 per million tokens, with no output or cache charges. OpenAI API docs
• Claimed Typed answers come about 10x faster than through the Responses API. OpenAI API docs