OpenAI Launches Decisions API to Challenge TypeSafe's Jev — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI made its Decisions API widely available to all customers, reconfiguring ChatGPT Luna to deliver responses in 100–180 milliseconds, down from seconds.
- TypeSafe created the decision model category with its Jev model, recently achieving a $7.6 billion valuation and releasing an updated version with image and smart-if support.
- Nikunj Handa, OpenAI’s API product lead, said engineers internally identified 'a million use cases' for decision models after seeing Jev, prompting rapid development of their own version.
- Public benchmarks like the AIM System One Index and Artificial Analysis Decision Index rank Liquid’s D1 Ultra and Base Lab’s Elephant-1 ahead of OpenAI’s model, which placed eighth and fourteenth respectively.
- TypeSafe’s Jev model leads in real-world usage on OpenRouter and topped the JevMan benchmark, where models play PacMan by making split-second navigation decisions.
- Speak, a language tutor AI, cut response latency from 1.4 seconds to 0.13 seconds using the Decisions API, while Box, Datadog, and Ramp also replaced LLMs with higher throughput.
Why it matters: Decision models are becoming a distinct and measurable category of AI infrastructure, with OpenAI now competing directly against startups like TypeSafe. The shift enables real-time, low-cost automation in software workflows—cutting latency by over 90% in cases like Speak’s tutors—and signals a structural move beyond general LLMs toward specialized, fast inference engines that could reshape how AI is embedded in applications.
Ask SkimNews
