OpenAI makes its Jev competitor widely available as decision model competition heats up
techcrunch ·
OpenAI has officially released its Decisions API to the general public, entering a rapidly growing market for fast and cost-effective decision-making models. This move follows the success of TypeSafe’s Jev model, which demonstrated that specialized 'System One' models could replace traditional large language models in specific workflows with significantly lower latency and expense. OpenAI adapted its ChatGPT Luna model to create this new offering, achieving inference speeds between 100 and 180 milliseconds compared to the previous one-to-three-second range. Early adopters like Box, Datadog, Ramp, and Speak have reported substantial improvements in throughput and response times, with Speak reducing action intervals from 1.4 seconds to just 0.13 seconds. While benchmarks show mixed results against competitors like Liquid’s D1 Ultra and Base Lab’s Elephant-1, OpenAI highlights the multi-modal capabilities of its Luna-based solution as a key advantage. The company is also exploring even faster speeds, potentially aiming for 60 milliseconds to enable real-time applications such as per-frame video analysis. As these models become cheaper and more ubiquitous, they are unlocking new use cases that were previously impractical due to speed or cost constraints.
techcrunch