Jev Docs

Release notes for the official Python SDK, from the 14 September 2026 public release through the Pydantic migration and the HTTP/2 extra.

Release notes for the official JavaScript and TypeScript SDK: the 11 September 2026 public release and the Score criteria breaking change.

RetryPolicy for the Python SDK: attempt count, exponential backoff with jitter, which HTTP statuses retry, and a total time budget.

The Python SDK error hierarchy: HTTP status classes, rate-limit wait hints, connection and timeout errors, and response-body validation failures.

AsyncTypeSafeClient: async system_one calls, model listing, per-call retry and timeout overrides, and an optional Pydantic response model.

How the Python SDK types questions: Noul, Choice, and Score as objects or dictionaries, including optional JSON instructions and criteria.

The Python response models: SystemOneResponse groups answers by id, and records the model name, token usage, and request id.

Default settings and environment variable names for the Python SDK, including the API root, model alias, and per-request timeout.

TypeSafeClient, the synchronous Python client: ask questions, list models, and set the key, model, retry policy, and HTTP timeout.

Shared Python SDK types: JSONValue, a nested JSON-like value that may be None, and descriptions built from strings or JSON values.

OpenRouter's guide for calling Jev with the official JavaScript or Python SDK by changing the base URL and using an OpenRouter key.

Databricks docs for the System One route on Unity Gateway. The example model is system.ai.openjev-qwen35-4b, with noul, choice, and score questions.

An independent wiki on TypeSafe's System One model: primitives, pricing, latency bands, console playground links, and ecosystem pointers.

A third-party developer reference for the Jev API: primitives, quickstart curl, SDK pointers, and published latency and pricing figures.

Rust docs for typesafe-jev 0.2.0, a blocking client that sends one state and a set of Choice, Score, and Noul questions and prices input tokens at list rate.

The index of TypeSafe's worked examples, from parallel questions and reranking to guardrails, date extraction, and self-consistency, each with datasets and measured results.

Guidance on how Jev fits coding agents: it is not a drop-in agent model, so the page routes readers to routing, rubric scoring, and truth checks instead.

The hub for TypeSafe's legal and policy documents, plus the data commitments attached to the API.

The generated API reference for the official JavaScript SDK, listing every exported class, interface, function, and type.

The generated API reference for the official Python SDK, split into sync and async clients, question and response types, retries, exceptions, and constants.

The full usage guide for the Python SDK, covering typed calls, model selection, retries, error handling, logging, environment variables, and forward compatibility.

The reference site for the community Spring AI integration: a Java client plus judges, guardrails, RAG filters, and tool search built on Jev.

The docs site for an unofficial Go client that wraps /v1/systemone and /v1/models with typed question and answer types.

TypeSafe AI's product site for Jev, its first System One Model: typed decisions with calibrated confidence, performance and pricing claims, a FAQ, and links to the docs, console, workflow evals, and launch post.

The raw HTTP contract behind every SDK: POST a state plus typed noul, choice, and score questions to /v1/systemone, and get one answer per question, with error codes and retry guidance.

The model card for jev-1.13.0: $42 per billion input tokens with free output, a 64k context, 250k tokens per second, and how to list the models your account can call.

How to shape what Jev evaluates: a string, a named object, or an array of messages. Every question in a request sees the same state and is evaluated independently.

The idea behind Jev: models that return typed decisions and probabilities instead of generated text, named after Kahneman's fast-thinking System 1, with confidence to route around uncertainty.

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.

A maintained list of jev-1.13's known failure modes, literal reading, unreliable counting, dates as text, indirection, context rot, and contradictory criteria, each with a guardrail.

The official SDK hub: Python and JavaScript/TypeScript clients with typed questions and answers plus automatic retries, and the HTTP API for every other language.

Quickstart for @typesafe-ai/sdk on Node 20+: answer types inferred from your questions, a choice() helper, and one systemOne call to classify a support ticket.

Usage guide for typesafe-sdk: sync and async clients, uv and pip installs, and one request mixing Noul, Choice, and Score questions with answers grouped by type.

The canonical introduction: Jev is TypeSafe's flagship model and the first System One model. Send state and typed questions, get structured answers your code can use directly.

Everything you need to make your first request: install an SDK, get an API key, and send state plus typed questions.