jev is insane 🤯 Here is Jev playing subway surfers at super human speed, and also playing 50 games at once. cost less than a cent to do this run. Jev does not replace llms like astra or fable, but opens up an entirely new world of capabilities.
Evaluation & TrainingReinforcement learning from human feedback
RLHF
The human-preference training method that turned pretrained models into chatbots, and the method TypeSafe deliberately did not use.
- Category
- Evaluation & Training
- Also known as
- Reinforcement learning from human feedback
- Related terms
- 5
- Directory entries
- 1
- Docs
- docs.typesafe.ai
- Added
- 2026-09-24
Definition
RLHF trains a model to produce responses people prefer. It powered InstructGPT and ChatGPT and was co-invented by Diogo Almeida, now a TypeSafe cofounder, but the primer argues preference optimization can reward sycophancy and confident-sounding hallucinations.
It remains a good fit for conversational models. TypeSafe's position is that production automation needs a different objective — constrained decisions and calibrated uncertainty.
Tagstraining
Related terms
Definitions that connect to this one.
RLCDTypeSafe's post-training method: reinforcement learning that optimizes for calibrated decisions and probabilities instead of generated text.Evaluation & TrainingRLVRTraining against automatically verifiable rewards, which produced strong reasoning models that are slower and more expensive.Evaluation & TrainingSycophancyThe tendency of preference-trained models to say what people want to hear, one of the reasons TypeSafe chose a calibration objective.Evaluation & TrainingMode droppingPreference optimization narrowing a model's output distribution toward a favored style and away from other valid outputs.Evaluation & TrainingTypeSafe AIThe company behind Jev and the System One API, including the official SDKs, cookbooks, console, and published workflow evals.Core Concepts
From the directory
Why TypeSafe trains decision models with RLCD instead of RLHF: calibrated probabilities where 0.2 outcomes happen about 20% of the time, and the case for machine-to-machine automation.
Sites & GuidesDocs#official#docs#evaluation
Official
From the community
Posts from builders shipping with Jev right now.
Jev plays Subway Surfers
Agentic browsing in Chrome
I built a Chrome extension for agentic browsing using Jev by @typesafeai, fx.sh including AI Gateway by @vercel. Now agents can browse, click, and interact with websites directly in your browser. Cost effective and fassst. Decision-making by Jev.
@CompleteSkeptic
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x
724 competitor ads, broken down
jev is INSANE. in 40 seconds it broke down 724 live ads from 37 brands. every hook. every format. offer. cta. awareness stage. landing page mismatch. used 9 cents of tokens. (will be avail in @stealads + mcp)
A chat bot with no LLM
I built a chat bot with jev, no LLM at all! Responses are instant, no hallucinations. I hooked it up to web search, wikipedia, weather, todoist and home assistant. Jev decides what tool to call and what args to use based on the prompt. Instant answers cite sources as well!
3,282 posts, eight questions each
I gave Jev 3,282 of my X posts across 100M views and asked it to find what actually works for growth. 4,252,330 tokens $0.1282 for the full 8m 34s run! Each post got 8 questions about the topic, hook, tone, whether it teaches something, etc. How-to posts got 150 median likes Show more
Jev repositories worth a look, in Japanese
やぁ！兄弟たち！ Jevに関するGitHubの実用性と発展性がありそうなリポジトリをまとめたよ！ やはり、高速判断を要するComputerUseや完全自動トレードなんかに対しての活用が多い印象だね！ Jevは公式のウェイトリストも1日ほどで承認されるけど、待たなくてもVercel AI GatewayからModel: Show more