OpenAI Broadly Releases Decisions API to Enter Fast Intelligence Market
OpenAI has made its Decisions API widely available, joining the growing market for fast, low-cost intelligence models that can act as super classifiers.
OpenAI has officially made its Decisions API available to all customers, bringing its own competitor into the expanding market for rapid, economical intelligence models. This segment was recently energized by startups like TypeSafe, whose Jev model created a category for deep learning systems acting as super classifiers. Given a set of choices, these models can quickly make selections and provide a confidence rating.
Software developers quickly realized that these System One models could often substitute traditional large language models in various workflows, operating faster and at a much lower cost. Nikunj Handa, the product lead for OpenAI’s APIs, noted that his team saw Jev and immediately recognized numerous potential applications. Rather than building a dedicated decision model from scratch, OpenAI reconfigured its ChatGPT Luna model to accelerate inference and reduce expenses.
The main breakthrough of the Decisions API is its speed. According to Handa, response times dropped from one to three seconds down to between 100 and 180 milliseconds. Initially unveiled at the company’s developer day on September 29, the Decisions API reached general availability on a Thursday, adding new voice and visual modalities while optimizing global inference latency.
Early trials included companies such as Box, Datadog, Ramp, and Speak, where engineers used the API to swap out standard LLMs for higher throughput. Andrew Hsu, CTO at language tutor builder Speak, reported that the Decisions API reduced action intervals from 1.4 seconds to 0.13 seconds. OpenAI also integrated the technology to boost the responsiveness of its Dots voice assistant.
What it means for developers
For developers seeking efficient ways to integrate AI, these fast decision models offer a compelling alternative to heavier LLMs for specific classification tasks. Because speed is measured in milliseconds, engineers can handle high-frequency workflows far more affordably. For those looking to experiment, developers can try top AI models cheaply through one API at https://apixoai.online, simplifying access across multiple providers without managing separate keys.
While benchmarks offer guidance, defining what makes a strong decision model remains nuanced. Public indices such as the AIM System One Index and Artificial Analysis Decision Index highlight various competitors like Liquid’s D1 Ultra and Base Lab’s Elephant-1, alongside TypeSafe’s Jev and OpenAI’s entry. Furthermore, unique evaluations like JevMan—where decision models play PacMan—test a system's ability to handle a nonstop stream of quick, high-stakes choices.
OpenAI believes its model holds advantages in multi-modality, particularly with image classification. Handa highlighted that Luna is exceptionally powerful in that area, and noted that future speed increases might drop latency even further, potentially enabling API calls for every single frame in a video.
Source: OpenAI makes its Jev competitor widely available as decision model competition heats up — TechCrunch AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
