OpenAI Decisions API Now Generally Available at $0.10 Per Million Tokens with No Output Charges

OpenAI has made its Decisions API generally available, entering a fast-growing market for lightweight, high-speed AI decision models. The API is powered by GPT-6 Luna and delivers responses up to ten times faster than requests sent through OpenAI's conventional Responses API. It supports yes/no predicates, multiple-choice decisions, and numerical scoring over text and images, making it suited for autonomous agents that require rapid, structured judgments at scale.

Pricing is set at $0.10 per million input tokens with no separate output-token charge, and cached input costs just $0.01 per million tokens. The absence of output-token fees marks a notable departure from OpenAI's standard model pricing and is designed to lower costs for workflows involving large numbers of automated calls.

The launch puts OpenAI in direct competition with Jev, a decision model from TypeSafe, a startup founded by a former OpenAI researcher that recently reached a $7.6 billion valuation. Microsoft, Cloudflare, and Perplexity have also entered the decision-model space, and open-weight alternatives are available for organizations that cannot send sensitive data to external APIs. The competitive pressure has accelerated an industry-wide shift in focus from benchmark performance toward speed and cost efficiency as AI moves deeper into automated, multi-step enterprise workflows.