Jev is just a trained model afawk., the (incomplete) papers the Laya author uses to claim they had the idea first describe RAG systems, orchestrators, and related built on LLM APIs. Laya came out after Jev and appears to be vibe coded to look more like Jev, then the claim of "idea theft" was made with the new blog post and code base, pointing back to papers that describe nothing like Jev.

Stepping back from Laya, there are dozens of Jev-a-likes now, most built on LLMs, whereas it is speculated that Jev likely uses transformers, but is not LLM like by generating a token at a time. Most likely they are reaching into the logits to get the details and assembling the json object deterministically. This is what the various LLM spins are doing, which is probably close but not the same as Typesafe.

This is the main PR (merged) for using DiffusionGemma (diffusion generally) for Jev-a-likeness in vLLM. https://github.com/vllm-project/vllm/pull/57250