What is a System One (System 1) model? Decision and extraction models
A System One (System 1) model returns typed, calibrated answers instead of text. Two classes now exist: decision models like Jev and extraction models like Clairic.
Updated
Guides to System One models and Jev
Explainers in reading order: what a System One model is, its two classes, decision and extraction models, what Jev returns, how it differs from a language model, and what the vendor's numbers do and do not prove. Every factual claim carries a source.
All guides
What is a decision model? AI decision models explained
A decision model answers typed questions about content with calibrated probabilities, not text. The question types, the vendors' names, and how it differs from an LLM.
Updated
What is a System One extraction model? Extraction models explained
An extraction model reads input and returns typed values, such as spans, numbers, dates and forecasts, with probabilities. How it differs from decision models and LLMs.
Updated
What is Jev AI? TypeSafe's decision model explained
Jev is TypeSafe AI's decision model. It answers typed questions about text with calibrated probabilities, and it is not an LLM, because it writes nothing.
Updated
System One models vs LLMs
System One models return typed decisions or values; LLMs return text. A side-by-side on output, latency, cost, hallucination and when each one wins.
Updated
Choice, Score and Noul: the three primitives
Choice, Score and Noul are the three question types a decision model such as Jev answers. Concrete examples, their limits, and how confidence is derived.
Updated
Clairic field types: text, number, date, list, estimate and forecast
The nine field types Cloudflare's Clairic extraction model returns, what each answer holds, and how p_null, guess, confidence and null_threshold work.
Updated
How to build with System One decision models
State in, typed questions out. Code owns the thresholds and side effects. The four official patterns, and the jaggedness list as design constraints.
Updated
RLCD explained: Reinforcement Learning for Calibrated Decisions
RLCD is TypeSafe's training method for Jev. What Reinforcement Learning for Calibrated Decisions means, how it differs from RLHF and RLVR, and what is not public.
Updated
Is Jev just a zero-shot classifier?
Engineers called Jev a zero-shot classification model within hours of launch. What the comparison gets right, what it misses, and what is actually new.
Updated
Jev speed and pricing
Jev costs $0.042 per million input tokens with free output. TypeSafe claims 70ms to 500ms and 40x to 200x. What those numbers cover.
Updated
How to get Jev API access
How to get Jev API access: open sign-up at the TypeSafe console, the key page, the TYPESAFE_API_KEY env var, SDK installs, and a first working call.
Updated
Jev architecture: parameters, encoder or decoder, and no paper yet
TypeSafe has not published Jev's parameter count, base architecture or weights. Here is the public record and what the community has guessed from it.
Updated
Jev 1.14: what changed, when jev-latest moves, and the danger zone
Jev 1.14 shipped on 14 October 2026 as jev-preview. What changed, when jev-latest switches, the opt-in image, box and long-context features, and deprecation.
Updated
Jev alternatives: hosted, open-source and local options compared
The realistic Jev alternatives in October 2026: other hosted APIs, open models you host, community clones and plain classifiers, with the benchmarks behind each.
Updated
Jev benchmarks: every public evaluation in one place
Every public Jev benchmark we could verify: TypeSafe's own evals, independent tests against Opus, Haiku, GPT and Laya, and what they do and don't show.
Updated
Jev game demos: Doom, Pokemon, Mario, drones and more
Every Jev game and real-time control demo in the directory: Doom, Pokemon, StarCraft, Mario, Tetris, Snake, chess, poker, drones, robot arms and driving.
Updated