OpenAI has introduced a new Decisions API designed to streamline evaluations for text and images. According to a report by The Decoder, the system performs classification tasks roughly ten times faster than the existing Responses API. The service delivers outputs such as yes or no probabilities, category selections, or scale ratings. Pricing is set at $0.10 per million input tokens. Alongside this software release, the company reduced its paid API tiers from five down to three.
API Structure And Pricing Context
Large language models are traditionally built to generate open-ended prose, which can introduce unnecessary latency and cost when an application only requires a binary choice or a category selection. By restricting outputs to structured choices like yes, no, or a specific category pick, specialized classification endpoints typically optimize both throughput and computational overhead. OpenAI’s reduction of paid tiers alongside this rollout points toward an ongoing effort to simplify its commercial developer offerings.
Unknowns And Limitations
Despite the initial details, significant information regarding the Decisions API remains unaddressed. The announcement did not specify technical benchmarks or accuracy metrics compared to larger foundational models. Furthermore, it remains unknown which specific underlying models power the endpoint, how rate limits will be applied across the newly consolidated paid tiers, or when broader availability for all developer accounts will be fully realized. Developers will need to monitor documentation updates for these operational specifics.
Adoption will depend on how reliably the new endpoint handles nuanced inputs compared to standard chat completions.
Lead image: Daniil Komov / Pexels
