GoKawiil
Tech News
Diogo Almeida, a former OpenAI researcher who helped invent RLHF and build ChatGPT, has released a new transformer-based model called Jev through his startup TypeSafe AI. Unlike large language models, Jev doesn't generate text—it produces calibrated probability outputs, making it faster, cheaper and immune to hallucination. Developers have shown strong demand, with the company briefly unable to keep up with API traffic.
A Show HN project puts four AI models in a head-to-head Pong match, running each in its own lane where the ball's speed depends entirely on that model's response time. The setup contrasts Jev, a typed-decision model, against chat-based models like GPT-5.6 and Claude Haiku, using Pong specifically because it exposes how chat models struggle with fast, continuous decision-making.
TypeSafe has open-sourced Jev Ultrafast, a browser automation agent that reads a page's elements as an indexed table and picks an operation-target pair in a single network round trip, calling a small language model only when text needs to be generated. In a demo, it booked a Zürich-to-London flight search on Google Flights from a single natural-language goal in 7.1 seconds, including page load waits. The code, along with a local inspector interface for watching element probabilities and executed actions, is available on GitHub.
A developer built an open-source model that mimics TypeSafe's proprietary Jev system, which scores a variable list of text options in a single pass rather than generating answers word-by-word. The project, released without official documentation from TypeSafe, includes working demos where the same attention-based scoring head selects Doom combat moves and chess moves from image inputs, along with checkpoints, code and a rendering pipeline to visualize decision traces.
TypeSafe AI, founded by former OpenAI researcher Diogo Almeida, has released Jev in early access, a new type of AI model designed to make fast structured decisions rather than generate text. The company says Jev matches existing large language models on decision-making tasks while running 20-200 times faster and 40-400 times cheaper, using a new architecture and a training method called Reinforcement Learning for Calibrated Decisions.