Jev-like decision models are flooding the AI market, after researcher claims TypeSafe copied their research
Jev and the flood of AI decision models — TypeSafe allegedly copied researcher's architecture but got copied itself
TypeSafe's Jev has sparked a wave of similar AI models across the market, following claims that its creator, Nandakishor Mukkunnoth, had developed the underlying concepts a year earlier. Jev is an API-driven decision model that processes input data and user questions, returning probabilities for specific answers. This approach bypasses the usual hurdles in prompting large language models (LLMs) to deliver direct answers. Unlike LLMs, Jev provides rapid, parallel responses and is considerably more cost-effective.
Mukkunnoth published his research on probability estimation model, SalesRLAgent, on arXiv in March 2025, and later a September paper that aligns with Jev's operation. Despite his earlier work, Mukkunnoth expressed frustration over being overlooked as TypeSafe launched a general-purpose decision engine, Jev, as a hosted product. Unperturbed, Mukkunnoth released an open-source alternative, Laya, shortly after.
Following Jev's release, numerous players in the AI industry launched their versions, including Microsoft Decision, OpenAI Decisions API, and Fastino Glide. Other options, such as Cloudflare Clef and Perplexity Decider, made their services available with open weights, allowing users to run them on their own hardware. Open-source alternatives, such as Amazon's Strands Decider 2B, Kev, Von, Jeff, jevlike, SemIf, OpenJev, NanoJev, and Ollaya, have also emerged.
Many of these projects share a common foundation of taking lightweight models, often based on Alibaba's open-source Qwen, and adding layers to customize input, output, and decision-making.
Written by urgent.news from Tom's Hardware's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
