OpenAI Launches Decisions API Public Beta on GPT-6 Luna With Faster Responses
The new API provides typed probabilities and choices at significantly higher speeds with a distinct token-based pricing model for enterprise developers.
Key points
- The Decisions API is now available in public beta on the GPT-6 Luna model.
- Response speeds are approximately ten times faster than the existing Responses API.
- Billing is set at $0.10 per million input tokens with no charges for output.
OpenAI has launched its Decisions API into public beta, deploying the service on its GPT-6 Luna model. According to a report by MarkTechPost, this new interface is designed to return typed probabilities, choices, and scores significantly faster than the company’s previously available Responses API.
The primary technical advantage of the Decisions API is its speed, with MarkTechPost noting that it delivers results approximately ten times faster than the standard Responses API. This performance increase allows for quicker processing of structured data requirements. The API specifically handles typed outputs, including probabilities and choice scores, which are critical for applications requiring deterministic or structured responses rather than open-ended text generation.
Context and Technical Specifications
The rollout marks a shift in how OpenAI structures its application programming interfaces for specific decision-making tasks. By focusing on typed answers, the Decisions API targets use cases where precise data structures are required. The underlying model powering this beta release is identified as GPT-6 Luna in the source material. This naming convention suggests a specific iteration or variant of OpenAI’s generative models, distinct from previous general-purpose releases.
The speed improvement is relative to the Responses API, which serves as the current benchmark for general interaction. A tenfold increase in processing speed can significantly reduce latency in automated workflows. This capability allows developers to integrate more complex decision logic into their applications without the traditional delays associated with large language model inference. The focus on scores and probabilities indicates a move toward more quantifiable AI outputs.
Who Is Affected
Developers and enterprises integrating OpenAI models into their infrastructure are the primary audience for this public beta. The pricing structure is designed to be predictable for high-volume input scenarios. According to MarkTechPost, the cost is fixed at $0.10 per million input tokens. This billing model differs from traditional usage-based pricing that often charges for both input and output tokens.
The absence of output charges is a notable feature for applications that generate extensive structured data. Users who previously paid for both the prompt and the resulting text will see a change in cost dynamics. This model encourages the use of the API for tasks where the input volume is the primary driver of cost, such as batch processing or high-frequency decision loops. Organizations currently using the Responses API may evaluate the Decisions API for performance-sensitive tasks.
What should you do?
Security and technology professionals should monitor the stability and behavior of the new API during the beta phase. Evaluate whether the tenfold speed increase aligns with your application’s latency requirements. Review the billing implications of the input-only pricing model against your current token usage patterns. Test the typed probability outputs to ensure they meet your data structure needs.
Background: Large language models
Large language models are a specific type of neural network trained on text. Generative AI is the broader umbrella term for any system that creates new content, including images and code. Not all generative AI uses large language models, but most modern text generators do.
What to do and how to stay safe: OpenAI
- Review current API usage patterns to determine if the input-only pricing model offers cost savings over the Responses API.
- Test the Decisions API in a sandbox environment to verify the accuracy of typed probabilities and choice scores.
- Monitor latency improvements to assess if the tenfold speed increase impacts your application’s user experience.
- Ensure your data handling procedures account for the structured nature of the API outputs during the beta period.
Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What model powers the new OpenAI Decisions API?
The Decisions API is powered by the GPT-6 Luna model, according to MarkTechPost.
How does the speed of the Decisions API compare to other OpenAI APIs?
It is approximately ten times faster than the Responses API for returning typed answers.
What is the billing structure for the Decisions API?
The API costs $0.10 per million input tokens, with no charges for output tokens.
