From OpenAI · · 1m 30s
Build responsive apps with Decisions API
OpenAI launches Decisions API for faster, cheaper app decisions.
In brief
OpenAI launches Decisions API for faster, cheaper app decisions. OpenAI's Decisions API helps apps make near real-time decisions 10x faster with GPT-6 Luna, supporting up to 400 questions per request. Originally reported by OpenAI.
Decisions API launch
OpenAI makes Decisions API generally available for faster app decisions.
GPT-6 Luna integration
Decisions API uses GPT-6 Luna to handle questions with bounded answers up to 10x faster.
Key improvements
Added native GPT-Live integration, prompt caching, and inference residency in US and EU.
New capabilities
Supports up to 400 questions per request, twice the previous limit with native json inputs.
Use cases
Teams use Decisions for transaction categorization, coverage classification, and cost attribution.
Customer examples
Box reduces route selection time by 5x with same accuracy as Responses API.
Datadog integration
Datadog uses Decisions to score agent-generated commands 40x faster with comparable accuracy.
Ramp cost savings
Ramp cuts AI cost attribution to 30% cheaper with same accuracy as GPT-6 Luna.
Sierra performance
Sierra achieves 90% lower cost per call with 3x faster decisions in live conversations.
Speak learning experience
Speak reduces time to next action from 1.4s to 0.13s for language learners.
Answer types
Decisions API supports probability, choice, and score answer types for different use cases.
GPT-Live integration
Decisions API integrates with GPT-Live for real-time voice experiences and expressions.
Pricing details
Decisions API costs $0.10 per million input tokens with cached inputs at $0.01 per million tokens.
The short version
OpenAI's Decisions API helps apps make near real-time decisions 10x faster with GPT-6 Luna, supporting up to 400 questions per request.
