SDKs & clients
vLLM Jev runs open Jev-style decision models on your own hardware: one command downloads a Hugging Face checkpoint and serves Choice, Noul and Score questions over HTTP.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
It targets Linux GPUs and Huawei Ascend chips, with a preview for Apple Silicon Macs and an experimental Windows CPU mode for 24 checked text models. Valen and Clef also accept short videos. The project's own speed figures come from A800 and Ascend 910B3 runs, and its notes warn that Kev-0.8B can change its selected label under concurrent batching.
You can have a developer run this tool on your own computer or server to host open AI decision models. Your app sends a question with a list of answers over your local network. The server picks the best choice and returns a probability score for each option.
A developer can start by running a supported model like Open-Jev-2B on a Linux computer with a dedicated graphics card. Be aware that models like Kev-0.8B can occasionally flip their chosen answers when handling several incoming questions at the exact same moment.