OpenAI Decisions API

★★★★☆4.3 / 5 editorial Paid Public beta at $0.10 per 1M input tokens; no charges for output or cache tokens. Runs on GPT-6 Luna only.

Typed answers your code can act on — probabilities, choices, and scores, not prose.

Visit Site ♡ 2,673 saves

#api#developers#agents#classification#routing

What is OpenAI Decisions API?

OpenAI pushed the Decisions API into public beta on October 6, 2026, and it is a deliberately narrow tool: instead of asking a model for a paragraph and parsing out a verdict, developers send text, images, or both to a dedicated POST /v1/decisions endpoint and get back typed answers their code can branch on. It was introduced in limited preview at DevDay 2026 in late September before opening to everyone.

Each request carries named questions in one of three types: a predicate returning a 0-to-1 probability, a choice from a fixed list with confidence scoring, or a score against ordered severity levels. OpenAI says it answers about ten times faster than asking the same question through the Responses API — a company claim, not an independently benchmarked one. It runs exclusively on GPT-6 Luna, priced at $0.10 per million input tokens with zero output or cache charges, and general availability is expected in the coming weeks.

Key Features

Three typed answer shapes: probability, fixed-list choice with confidence, and severity score.
Dedicated POST /v1/decisions endpoint — no JSON parsing or regexing prose required.
Accepts text, images, or both in a single request, with multiple independent questions allowed.
$0.10 per 1M input tokens with no output or cache charges — cheap for high-volume classification.
Powered exclusively by GPT-6 Luna; general availability expected in the coming weeks.

Pros

Purpose-built for the classification and routing work that currently wastes prose-model calls.
Pricing is genuinely aggressive: input-only billing cuts costs for high-volume decision workloads.
Zero Data Retention and HIPAA eligibility make it a fit for sensitive gating flows.

Cons

Public beta: accuracy and reliability figures at scale are not independently verified.
Locked to GPT-6 Luna and OpenAI-hosted only — no open weights or self-hosting.
The 10x speed claim is OpenAI's own number with no published benchmark behind it.

Pricing

Model: Paid
Public beta at $0.10 per 1M input tokens; no charges for output or cache tokens. Runs on GPT-6 Luna only.

Adding Up Your AI Bill?

If you pay for OpenAI Decisions API alongside other AI subscriptions, the monthly total adds up quietly. Tick every tool you pay for in our free AI Tool Cost Calculator to see your real monthly and yearly bill — and spot overlaps you can cut. Still choosing? Start with our guide on how to choose the right AI tool.