OpenAI makes its Jev competitor widely available as decision model competition heats up

Key Points:

• OpenAI has launched its Decisions API model publicly, aiming to compete in the fast, cost-effective decision-making AI market pioneered by startups like TypeSafe, whose Jev model popularized deep learning super classifiers.
• OpenAI adapted its ChatGPT Luna model to create the Decisions API, significantly improving speed by reducing inference time from 1-3 seconds to around 100-180 milliseconds, while lowering costs without specialized training.
• The Decisions API, announced on September 29 and available to all users, supports new voice and visual modalities and offers lower latency globally by geographically targeting inference resources.
• Early adopters such as Box, Datadog, Ramp, and Speak have reported substantial performance gains, with Speak’s CTO noting a reduction in action response time from 1.4 seconds to 0.13 seconds using the API.
• While benchmarks vary, TypeSafe’s Jev model remains competitive, topping some rankings like the JevMan PacMan benchmark, but OpenAI’s model stands out in multi-modal capabilities, particularly in image classification and calibration.