TypeSafe AI's flagship model and the first System One model: typed decisions and calibrated probabilities instead of generated text.
Jev is the model you call through TypeSafe's API. It evaluates a state against typed questions and returns one structured answer per question — a choice, a score, or a probability of yes — so code can branch on the result without parsing prose.
TypeSafe ships it under versioned IDs such as jev-1.13.0, with jev-latest and jev-preview as aliases. The docs describe it as fast and calibrated on common-sense judgment, with known jagged edges around counting, dates, and numeric precision.
Definitions that connect to this one.
Posts from builders shipping with Jev right now.
dmx.to reshapes the timeline in plain language
the next version of dmx.to will let you fully customize the timeline both with dumb rules (how did i not think of these until now lol) and smart rules using Jev (BYOK also supports vercel gateway) 🤓 you can filter 3 lvls of usefulness, tweet types, topics, etc
GPT and Jev compared live
Built a live tool using @typesafeai JEV ! See live comparison btwn GPT & JEV at gptvsjev.vercel.app Get benchmarks like Money Spent ! Time Taken ! Tokens Used!
LitJev: Jev-style decisions on any local LLM
开源了 LitJev：我们对 Jev 类型化决策 API 实现方式的猜想，可接入任意现成 LLM。 小测试：Qwen3.8-27B，10 道 MMLU-Pro 一次完成，0.3 s。比让模型只生成答案快 2.9x；比生成完整 schema JSON 快更多（显然）。更多测试马上来。 非官方，基于公开信息。
Secret detection with Jev, tested
Built a quick Secret Detection Test with Jev by @typesafeai Results seem promising so far, ran it a few times and it's pretty consistent as well. github.com/teyhouse/jev-s…
Jev Driving Lab: a playground for System One models
@typesafeai (and System One models in general) have a lot of uses, and fun to play with. New paradigm open to public now. This is a self-driving sim I spun up to test Jev today. Structured sensor state in, typed driving decisions out, steer, brake, overtake, pedestrians, speed Show more
200 rows, 32 template families, and traps, judged in one call
First test with Jev (TypeSafe System One): 200 rows, 32 template families, traps — 'yes' rows with "we'll pass", 'no' rows with 'absolutely, happy to meet'. The single Jev question scored 100%. My 12 dimensions scored 98%. I had designed a task my own question could ace.
