TypeSafe
An AI infrastructure company specializing in the development of AI models purpose-built for decision-making, focused on providing low-latency, low-cost structured reasoning services for high-frequency, real-time decision scenarios.
Core Facts
Jev is a type of front-end model called a 'System 1 model' — fast, cheap, and structured, designed specifically for decision-making; it does not generate text but instead returns probability scores within a predefined set of options.
Decision-1 can be called in three ways: the AI SDK's decision API, the OpenAI-compatible Decisions API, and the TypeSafe-compatible API.
JEV is a decision model launched by TypeSafe
Timeline (last 90 days)
Pedantic, through TypeSafe, has released a decision model version that can be plugged directly into the agent loop.
TypeSafe's rule of thumb is: if a group of smart people could answer a question within a few seconds, a decision model should be used; if it requires thinking and composing a complex answer in words, keep using a large model.
Jev 1.14 was released in preview as jev-preview on October 14, and jev-latest still points to 1.13.
Jev is fully closed-source and only offers a hosted API (api.typesafe.ai), and can also be called via OpenRouter.
The three System One decision models share the same API: POST /v1/systemone, passing in state and a question mapping; this protocol is defined by TypeSafe, Jev is the first implementation, and Drex and Laya are wire-compatible.
The decision model behind Armor is TypeSafe's Jev, currently at version typesafe/jev-1.14, a structured decision model that returns typed answers with probabilities.
TypeSafe's Jev Router scored 0.6398 and 0.8523 on two builds on the Forge 2.0 agentic coding benchmark, with a median of 0.746, ranking first.
TypeSafe released Jev on September 15, a new product positioned as a decision model rather than a chatbot.
Starting from TypeSafe's release of Jev in mid-September 2026, at least eight companies launched AI decision models within a little over a month, including Cloudflare's Clef, the AWS Strands team's Decider, Perplexity's pplx-decider, Liquid AI's d1 series, OpenAI's Decisions API and Microsoft-Decision-1.
TypeSafe's Jev is a hosted service and its underlying model has not been disclosed.
40 more timeline events
All Facts (20)
Jev is a type of front-end model called a 'System 1 model' — fast, cheap, and structured, designed specifically for decision-making; it does not generate text but instead returns probability scores within a predefined set of options.
75%VerifiedDecision-1 can be called in three ways: the AI SDK's decision API, the OpenAI-compatible Decisions API, and the TypeSafe-compatible API.
65%VerifiedJEV is a decision model launched by TypeSafe
65%VerifiedTypeSafe positions Jev as a 'System 1 model,' a concept inspired by the dual-system cognitive theory in Daniel Kahneman's Thinking, Fast and Slow.
65%VerifiedOpenAI launched a product called the Decisions API.
75%VerifiedJev, launched by TypeSafe, is an AI model that does not generate text but only outputs typed decisions with calibrated probabilities.
65%UnverifiedTypeSafe's rule of thumb is: if a group of smart people could answer a question within a few seconds, a decision model should be used; if it requires thinking and composing a complex answer in words, keep using a large model.
50%UnverifiedJev is fully closed-source and only offers a hosted API (api.typesafe.ai), and can also be called via OpenRouter.
50%UnverifiedThe three System One decision models share the same API: POST /v1/systemone, passing in state and a question mapping; this protocol is defined by TypeSafe, Jev is the first implementation, and Drex and Laya are wire-compatible.
50%UnverifiedThe decision model behind Armor is TypeSafe's Jev, currently at version typesafe/jev-1.14, a structured decision model that returns typed answers with probabilities.
50%UnverifiedTypeSafe's Jev is a hosted service and its underlying model has not been disclosed.
50%UnverifiedA decision model is a category of model that takes state (text, JSON, images) and a typed question and returns structured answers (probabilities, enumerated choices, ranking scores); it is named after TypeSafe's Jev and was released in September 2026.
50%UnverifiedDrex v2.0's API contract /v1/systemone is exactly the same as TypeSafe's Jev, so decision-calling code written for Jev can switch to Drex with almost no changes.
50%UnverifiedThe concepts of System One and System Two come from Daniel Kahneman's Thinking, Fast and Slow; TypeSafe borrows the concept to distinguish two model invocation paths within an agent.
50%UnverifiedTypeSafe is a model platform oriented toward large-model decision-making, whose core idea is to constrain the LLM's free-text output into typed structured results; Jev is the decision model runtime on this platform.
50%UnverifiedDuring this test, when connecting to Jev, the endpoint was set to https://api.typesafe.ai and the decision model used was jev-latest.
50%UnverifiedTypeSafe claims that Jev runs significantly faster than mainstream LLMs.
50%UnverifiedTypeSafe claims that Jev consumes far fewer tokens than mainstream LLMs.
50%UnverifiedJev is positioned by its developer TypeSafe as the first System 1 model, focused on classification decisions rather than text generation.
50%UnverifiedAccording to TypeSafe, depending on the situation, Jev may be 40 to 200 times faster than traditional LLMs.
50%