jev-ultrafast
Author browser-use
A public Python web-agent repository. GitHub's own description only calls it the fastest and cheapest web agent, and this page does not add how it calls Jev.
Open source
Public GitHub repositories whose description or README opening names Jev, TypeSafe, or System One. Checked on 2026-09-23.
These are other people's public repositories. JEV Hub does not maintain or host them and has not benchmarked them. A repository that mentions Jev is not necessarily part of TypeSafe; check its README.
Author browser-use
A public Python web-agent repository. GitHub's own description only calls it the fastest and cheapest web agent, and this page does not add how it calls Jev.
Author OpenByteInc
A trading system whose repository says it includes a Jev System One integration. That integration is not Jev itself.
Author tamaratran
A Claude Code plugin that uses one Jev request to score tool calls and results, drops or truncates what is stale, and leaves the rest verbatim.
Author TheoLeeCJ
Semantic conditionals from open models, aimed at a home GPU. The repository says it is independent and not affiliated with Jev or TypeSafe.
Author jaredpalmer
An independent family of small Jev-like decision models, built on Qwen3.5, that you can train and run yourself.
Author jev-chat
A phone chat copilot for WeChat, QQ, X, and Feishu: it reads the thread on screen, judges intent, risk, and whether to reply now, then offers three replies you can put in the input box without sending. It does not hook the app, change the install, call the chat API, or touch transfers, red packets, and payments; WeChat and QQ bubbles are read with accessibility, Feishu text with on-device OCR, and a manual screenshot can be read once. Keys stay on the phone, chat text is sent only to the endpoint you configure during analysis, the project has no server of its own, and a blank reply or vision endpoint inherits the judgment one, including an optional TypeSafe preset.
Author jarrodwatts
One AI trade decision on each Monad block, described by the repository as Jev on the Kuru MON-USDC market.
Author TianyuCodings
A nano replica of Jev with parallel decisions, dynamic candidates, and a training pipeline.
Author vinnylarouge
An independent starter model: given a passage and a set of options, one pass returns a probability for each option. TypeSafe has not published the design. A local visual demo is included.
Author yibie
A public list of Jev projects, integrations, and discussions.
Author reticlehq
Runtime perception for web and desktop apps that the repository calls Jev-style. It is not the Jev model.
Author Anil-matcha
A public list of use cases, patterns, prompts, and starter code for TypeSafe Jev, which the repository calls a System One model.
Author kitfunso
Agent memory with decay and consolidation. The repository says retrieval can optionally use a TypeSafe Jev reranker, and the memory layer is not Jev itself.
Author v-modal
A public list of tools the repository says were built for Jev, TypeSafe AI's System One model.
Author Sac-Y
Hands the next Computer Use click to Jev, which the README calls TypeSafe System One, choosing from on-screen text while a local policy gate holds sensitive actions.
Author devagrawal09
Staged code review and a local dashboard built with TypeSafe Jev.
Author jev-chat
A macOS floating panel for WeChat: when a message appears, it shows that line's intent, risk, and reply candidates, and filling only places the text in the input box without pressing send. It watches the screen only: no injection, no hook, and no decrypting the WeChat database, and the panel appears only while WeChat is in front. Judgment can use your own TypeSafe Jev key or a small local model offline; reply candidates go through an OpenAI-compatible or Anthropic-compatible endpoint you configure, or through local Ollama.