aimodelvs data as of 2026-10-08
home / models / Jev
TypeSafe AI · System One · decision model

Jev

Early accessEarly access (waitlist)

A non-generative 'System One' model that returns typed decisions with calibrated probabilities instead of text, at $0.042 per million input tokens.

Released2026-09-1523 days ago
Input$0.042per 1M tokens
Outputfreeper 1M tokens
Context windownot published
Max outputnot published
Knowledge cutoffnot published

What it is

Jev is TypeSafe AI's first model. You send application state plus a set of typed questions (choice, score, noul) and get back a chosen option or score with a probability and a confidence value. It does not write prose. TypeSafe prices input at $42 per billion tokens and output as free, and reports 70 to 500 ms end-to-end latency on its own evals.

Pricing

Input$0.042
Outputfree
Cached input (cache read)not published
Cache writenot published
Batch inputnot published
Batch outputnot published
Fast / priority inputnot published
Fast / priority outputnot published
  • Output is listed as free ('too cheap to meter').
  • TypeSafe's FAQ asks whether prices are temporary or subsidized without answering.
  • Early-access pricing; verify in the console before budgeting.

Capability highlights

  • Three question primitives: choice (pick one of up to 255 options), score (rate on a rubric), noul (0 to 1 probability a statement is true).
  • Every decision carries a confidence estimate; you set thresholds for when it acts and when it escalates.
  • TypeSafe claims 193.6x faster and 444.6x cheaper than LLMs on its own workflow evals (vendor claim, not independently verified).
  • Trained with what TypeSafe calls Reinforcement Learning for Calibrated Decisions.

Modalities: text (state + typed questions) → typed decision + probabilities + confidence. Non-generative; all questions evaluated in parallel against the same state. Context limits not published. Choice questions support up to 255 options.

How to use it

API: REST API via console.typesafe.ai; docs at docs.typesafe.ai. Third-party gateways (e.g. Netlify AI Gateway) list it too.

TypeSafe API console (early access)Netlify AI Gateway (per Netlify changelog)

Sources

Every number above comes from one of these pages, read on the date shown. Searched for: jev ai, typesafe jev, jev model.

  1. TypeSafe AI: Introducing System One models and Jev read 2026-10-08
    release date, pricing, latency, claims, cardinality
  2. TypeSafe docs: Jev and primitives read 2026-10-08
    question primitives, outputs
  3. TypeSafe AI home page read 2026-10-08
    early access, $42 per billion tokens