最后更新：2026-09-20 · 本页随新信源出现自动刷新
Jev 动态与更新
关于 Jev 值得知道的一切，按时间排列：官方公告、渠道上架、独立分析、社区里程碑。每条附来源链接；厂商自报数据均已标注。
Show HN: Will Jev pull the lever in the trolley problem?
Hacker News discussion — 3 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Understand any Bluesky account (Jev)
Hacker News discussion — 3 points, 2 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
What are you building with jev?
Hacker News discussion — 2 points, 6 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
JEV: Ranking 6K+ peer-company candidates in ~40 seconds
Hacker News discussion — 2 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Generating levels in real time with the Jev model
Hacker News discussion — 4 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Show HN: Jev-lint – semantic linter with plain English rules
Hacker News discussion — 5 points, 1 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Show HN: Jev-align, a CLI to calibrate Jev to your judgement
Hacker News discussion — 3 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Semantic end-to-end agent testing for iOS, Android, and web using Jev
Hacker News discussion — 2 points, 1 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Show HN: Jev, Fly Me to the Moon
Hacker News discussion — 2 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
How to classify, route, and score with Jev and AI SDK
Hacker News discussion — 2 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Inventor of ChatGPT and RLHF Launches Typesafe.ai
Hacker News discussion — 5 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
TypeSafe AI's Jev Is Not an LLM – and That May Be the Point
Hacker News discussion — 3 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Typesafe Manifesto - build prod not God
Hacker News discussion — 1 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Show HN: Explore 2D semantic space with the Jev model
Hacker News discussion — 2 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
TypeSafe AI's "Meaningful Intelligence"
Hacker News discussion — 1 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Show HN: Open-Source Alternative to TypeSafe.ai
Hacker News discussion — 2 points, 1 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Independent analyses urge caution on Jev benchmarks
Multiple third-party write-ups (The Tech Society, DataCamp, dev.to) compile Jev's numbers and flag that the 4-workflow benchmark is vendor-run, with consensus labels generated by GPT-6 Astra and Claude Fable 5.1 rather than ground truth. Accuracy on your own traffic remains the recommended test.
Jev is about to change the AI economy
Hacker News discussion — 4 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Typesafe AI
Hacker News discussion — 4 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
What's Next After RLHF? – Diogo Almeida, TypeSafe AI [video]
Hacker News discussion — 1 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
TypeSafe clears 140,000 users off the waitlist in 36 hours
After launch-day demand, TypeSafe reported moving 140k signups off the waitlist within roughly a day and a half, while asking new users to keep using the Vercel AI Gateway in the meantime. Production reports cited include a -71% cost and -90% wall-time reduction on a dispatch workflow, and 1,018 research papers classified for $0.08 at 256ms median latency.
Typesafe.ai Jev Open Source Alternative Qwen-2.5-1B-RLCD
Hacker News discussion — 2 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
TypeSafe AI debuts model for machines that plays Doom
Hacker News discussion — 1 points, 2 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Jev lands on Vercel AI Gateway
Vercel CEO Rauch Grover confirmed Jev is available through the Vercel AI Gateway under typesafe-ai/jev, reporting up to 18x faster p95 responses versus text-generating models on routing workloads. This becomes the fastest no-waitlist path to a Jev API key.
Cloudflare, OpenRouter and LangChain add Jev on day one
Cloudflare Workers AI lists typesafe/jev (32K context), OpenRouter lists typesafe/jev-1.13 (64K context, $0.042/1M input, $0 output), and LangChain publishes a guide for building evaluation harnesses with Jev via the langchain-typesafe package.
Typesafe.ai's System One Model
Hacker News discussion — 1 points, 0 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Introducing System One Models and Jev
Hacker News discussion — 1917 points, 502 comments. Community reaction and first-hand usage reports in the thread. 查看原文 →
Launch: TypeSafe introduces System One Models and Jev
TypeSafe AI launches Jev, a model that returns typed decisions (Choice, Score, Noul) with calibrated probabilities instead of generating text. Headline claims: 193.6x faster and 444.6x cheaper than frontier LLMs on System One workflows, $0.042 per 1M input tokens with output unmetered, and 0% schema/type errors by construction.