OpenAI Decisions API: 10x Faster Classification for $0.10

OpenAI introduced the Decisions API to streamline programmatic routing, document sorting, and visual inspection workflows. Instead of returning open-ended generated text, the endpoint restricts outputs to binary yes/no probabilities, categorical selections, or numerical scale ratings.

How much faster and cheaper is the Decisions API?

According to reports from The Decoder, the endpoint runs roughly 10 times faster than the standard Responses API. Input tokens cost $0.10 per million, while output tokens are entirely free. During the current public beta phase, the infrastructure supports only the gpt-6-luna model.

What changes for API tier management?

Alongside the release, OpenAI simplified its paid developer tiers from five levels down to three: Build, Launch, and Grow. Accounts automatically migrate upward when total credit purchases meet the next milestone, with monthly usage limits set at $500, $5,000, and $200,000 respectively.

Where does this fit into automated pipelines?

For developer workflows that require deterministic logic rather than conversational text, traditional LLM completions introduce unnecessary latency and parsing overhead. The Decisions API removes formatting errors by design, returning structured classifications that integrate directly into decision trees and conditional automation scripts without regex extraction.