Use the Decisions API when you need typed answers to a fixed set of classification or scoring questions about shared text or inline-image evidence. Use the Responses API for broader response generation, tool use, multimodal workflows, or multi-turn interactions. Decisions launched in beta on October 6, 2026, so treat it as a targeted option to test—not a drop-in replacement for Responses.
How the two APIs differ
| Decision factor | Decisions API | Responses API |
|---|---|---|
| Documented purpose | Evaluates ordered classification and scoring questions against shared input; answers follow question order. OpenAI API reference | General-purpose response interface for agent-like applications, including tools, multimodal support, and multi-turn interactions. OpenAI migration guide |
| Request input | Text or user messages containing text and inline image data URLs. The reference documents a maximum of 128 image parts per request. | Flexible string or message input, with native text-and-image support; broader workflows can use tools and typed items. |
| Result shape | Typed answers to predicate, choice, or score questions; an answer can also be a refusal. | An output array of typed items, which can include messages, function calls, and function-call outputs. |
| Tools and workflow | The documented request does not support tools or function calls as input. | Supports configured tools such as web search, file search, computer use, code interpreter, function calling, and remote MCPs. Availability and call handling depend on the model and integration. OpenAI tools guide |
| Conversation state | The reference describes shared input evaluated against ordered questions, not a conversation-state workflow. | Supports multi-turn interaction through the Conversations API or by chaining with previous_response_id. |
When to choose the Decisions API
Your output is a fixed evaluation
Decisions is designed for cases where the application already knows what it needs to ask about the evidence. Its documented endpoint, POST /decisions, evaluates ordered questions against shared input and returns answers in that same order. That makes it a fit for pipelines that need a compact, predefined result rather than a flexible generated response.
- Predicate: estimate how likely a statement about the input is true.
- Choice: select from options supplied with the question.
- Score: rate the input against supplied ordered levels.
Predicate answers contain a probability. Choice and score answers include confidence and probability information, and an individual question may return a refusal. See the Decisions API reference for the current request and response contract.
Your evidence fits the documented input contract
The reference documents text and inline image data URLs in user messages. It does not accept external image URLs or file IDs as substitutes for inline image data, and it lists files, audio, tools, function calls, function-call outputs, and item references as unsupported input. If your workflow relies on any of those, the documented Decisions request is not a direct fit.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →#1 Best Overall
When to choose the Responses API
You need more than a fixed set of answers
Responses is the more suitable starting point when an application needs flexible output, an agent-like sequence of actions, or tool use. OpenAI’s migration guide describes built-in web search, file search, computer use, code interpreter, remote MCPs, multi-turn interaction, and native text-and-image support. Check the tools guide for the current availability and integration requirements for a specific tool.
Your application carries state across turns
Responses documents two ways to continue an interaction: use the Conversations API or chain requests with previous_response_id. Its output is a typed array of items, so application code should handle the item types it may receive rather than assume every result is a single text string.
Responses are stored by default; the migration guide says storage can be disabled with store: false. Review the current data-handling requirements for your application before choosing storage behavior. OpenAI recommends Responses for new projects in its comparison with Chat Completions, but that is not a universal recommendation that Responses is always preferable to Decisions.
How to decide for your application
- Define the output contract. If each request should answer a known predicate, choose among explicit options, or assign a score level, prototype Decisions. If it needs flexible prose, tools, or multiple kinds of action, prototype Responses.
- Match the input to the endpoint. For Decisions, verify that your evidence can be sent as text or inline image data URLs. External image URLs, files, audio, tools, function calls, and item references are outside its documented input contract.
- Account for interaction state. If requests must continue a conversation, Responses documents Conversations and
previous_response_id. The Decisions reference documents ordered evaluation against shared input instead. - Verify beta availability and model support. The October 6, 2026 beta announcement names gpt-6-luna at launch. Check the live reference and changelog for current availability and supported models before implementation.
- Test representative traffic. Measure end-to-end latency, answer quality, refusal behavior, failure handling, and the operational cost of any extra orchestration with examples that resemble your workload.
- Plan for interface differences. When adapting an application, review request shape, typed outputs, state, storage defaults, structured-output configuration, and function-calling behavior. These are among the differences OpenAI identifies in its Responses migration guide.
What to make of the 10x speed claim
OpenAI’s October 6, 2026 changelog announcement says Decisions turns text and images into typed answers “10x faster than Responses.” The cited release entry does not state the benchmark setup, workload, measurement definition, or result distribution. Treat the figure as OpenAI’s launch claim—not a forecast of production latency—and test both endpoints on representative inputs before relying on it. OpenAI API changelog
Availability and what is established
Decisions was announced in beta on October 6, 2026, with gpt-6-luna named at launch. Beta availability, supported models, and the API contract may change. Responses is documented as OpenAI’s broader response interface; the recommendation for new projects in the migration guide applies specifically to its comparison with Chat Completions, not directly to Decisions. Check the current Decisions reference and changelog before selecting or shipping against the beta.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
