OpenAI's Jev killer is just Luna, sprinted out in weeks to chase a $7.6B rival

OpenAI opened its Decisions API to all customers on October 22 — a fast 'decision model' endpoint built to chase TypeSafe's Jev, which turned the startup into a $7.6 billion company. API product lead Nikunj Handa told TechCrunch his team saw Jev and thought 'we have like a million use cases for this,' then sprinted their own version. Rather than train a dedicated decision model, they reconfigured GPT-6 Luna to cut inference cost and latency — 'right now, it's just plain old Luna that gets the job done.'

The pitch is speed: decisions that took 1–3 seconds through OpenAI's normal APIs now return in 100–180 milliseconds, priced at $0.10 per million input tokens with no output charge. Early customers like Box, Datadog, Ramp, and Speak used it to swap LLMs out of workflows; Speak cut the gap between language-tutor actions from 1.4 seconds to 0.13. OpenAI also added voice and image modalities and pushed the endpoint to all customers on Thursday.