Recommended by Pro Trailblazer
TypeSafe AI
TypeSafe
Small typed judgments with a probability attached, fast and cheap enough that your code can ask on every request and branch on the answer.
$0.042 per 1M input tokens
About
TypeSafe sells API access to Jev, a System One model that answers narrow questions about application state with a typed result and a probability instead of generated text. You pass the state as JSON, ask one or more questions using the Choice, Score, or Noul primitives, and get back answers your code can branch on without parsing, retries, or a schema-repair step. Independent questions over the same state run in parallel in one request, and the probability on each answer lets you route the uncertain cases to a person or a reasoning model. Typed output guarantees the shape of the answer, not its correctness, so thresholds still need testing on your own data. We built the three TypeSafe and Claude Code demos on this site against it.
Highlights
- Three primitives cover most decisions: Choice picks one option from a set, Noul returns the probability a condition holds, and Score places an item on ordered levels
- Answers come back typed, so code consumes them directly with no JSON parsing, no malformed-output retries, and no prompt to keep in sync
- Every answer carries a probability, so you can set a threshold and hand the uncertain cases to a person or a reasoning model
- Ask several independent questions over the same state in one request and they run in parallel
- Billed per input token only, with output tokens free, at $0.042 per million input tokens for Jev 1.13
- Playground in the console, an HTTP API, Python and JavaScript SDKs, and a Claude Code agent skill
- Cookbooks for reranking, value extraction, citation checks, function calling, and hierarchical classification