Jev AI by TypeSafe: answers with probabilities, not chat

This site is an unofficial developer handbook for Jev AI and TypeSafe AI: API shapes, cost math, and when a decision model beats a chat LLM.

What Jev AI (TypeSafe) is for

Chat models generate words. The Jev AI model from TypeSafe returns typed decisions. The expensive part is not writing the reply. It is deciding what kind of work this is before you spend tokens on prose. System One is that first hop: classify, route, score, then hand off to code, a human, or a chat model.

Current public facts: context window 32,000 tokens; text-only input; model string jev-latest; input $0.042/M tokens with free output; claimed P50 around 0.23s. Vendor claims: 193.6x Faster and 444.6x Cheaper.

Three primitives

• Noul: yes/no with a probability of yes.
• Choice: pick one of up to 255 labeled options.
• Score: ordered scale (2-10 levels) with a distribution.

Keys in the questions map are your names. Answers return under the same keys. Keys do not participate in reasoning. Batch many questions in one request. The cookbook claims 13 questions in one call can be about 12.2× cheaper and 10× faster than asking them separately.

30-second triage example

One System One call can ask several questions at once. The JSON below is an illustrative example from the TypeSafe docs, not a measured production run.

{
  "state": "Customer email: My invoice charged twice this month and I cannot download the PDF.",
  "model": "jev-latest",
  "questions": {
    "is_support": {
      "type": "noul",
      "instructions": "Is this a customer support request that needs a human or bot reply?"
    },
    "department": {
      "type": "choice",
      "instructions": "Route this message to the best department.",
      "criteria": {
        "billing": "Payments, invoices, refunds, subscriptions",
        "technical": "Bugs, outages, API errors, product failures",
        "sales": "Pricing questions, upgrades, new purchases"
      }
    },
    "urgency": {
      "type": "score",
      "instructions": "How urgent is a reply?",
      "criteria": [
        "low",
        "medium",
        "high"
      ]
    }
  }
}

Hard numbers (official claims)

• 193.6x Faster · 444.6x Cheaper (TypeSafeAI 1.1 comparison on typesafe.ai)
• Input $42 / billion tokens. Output free.
• P50 latency about 0.23s. Context 32,000 tokens.
• ~300 input tokens ≈ $0.0000126 per request.

Where teams actually use it

• Support triage: noul for “needs human?”, choice for department, score for urgency, then confidence gates whether a bot replies or a ticket opens.
• Agent routing: let a chat model draft, then ask Jev AI whether the draft is ready, on-policy, or should escalate.
• Rubric judging: score or choice over fixed criteria when you need a distribution, not a paragraph of self-justification. OpenRouter traffic leaders include sel-jev-rubric-judge, Waura, mirasim, JeVinci, clara jev triage.

Naming rule on this site: say Jev AI or TypeSafe’s Jev AI in titles and first mentions. Prefer “Jev AI model” or “decision model” over bare “jev model”. The short form often collides with Japanese encephalitis virus and veterinary “animal model” results in English search.

How a production call usually looks

Keep API keys on a server. Send state (string, object, or array), set model to jev-latest, and pass a questions map. Prefer batching related questions in one HTTP request. Plan for 401 auth failure, 422 validation failure, 429 rate limit, and 529 overload. The last two should use exponential backoff. On OpenRouter, Jev AI rides the Decisions API, not chat completions.

instructions can also be structured data. Put the question in one field and keep references in sibling fields, then cite those fields with backticks inside the question text.

Architecture patterns worth copying

TypeSafe documents four composition patterns: speculative fan-out, confidence-gated routing, composite scoring, and intent routing. Most teams start with intent routing plus a confidence gate. Low confidence escalates. High confidence takes the cheap path.

Pairing models is normal. The chat LLM writes the customer-facing sentence. Jev AI decides whether that sentence should send, which template to use, or which queue owns the case.

What this handbook is not

We do not mirror the official docs page for page, we do not host TypeSafe branding, and we do not claim measured benchmarks unless we say so. Example JSON on this site is illustrative unless labeled otherwise.

The language button switches this handbook among English, Japanese, and Korean. Code samples stay as written.

Community cases

The case wall reprints 202 public posts: original text, a Chinese reference, and a link back to the post.

GitHub projects

The project directory lists public GitHub repos that ship Jev AI clients, agent guardrails, routers, and demos.

Tools on this site

• Pricing calculator: day and month cost at official rates.
• Request builder: copy-paste noul, choice, and score JSON.
• API quickstart: Python and a multi-question call.

A practical mental model

Treat Jev AI like a typed function, not a conversation partner. Prepare the state once, then ask several independent questions against that same state. Branch in code on the returned map. There is no assistant message to parse.

Batching keeps the decision layer cheap. Official cookbook numbers claim about 12.2× cheaper and 10× faster for 13 questions in one request versus asking them separately.

Confidence is the second control. Probability tells you how mass is distributed. Confidence tells you whether to act. Low confidence should escalate, not silently take the top option.

FAQ

Is Jev AI the same as TypeSafe AI?

No. TypeSafe AI is the company. Jev AI (System One) is their decision model. See the TypeSafe AI notes.

Is Jev Notes affiliated with TypeSafe AI?

Unofficial developer resource. Not affiliated with TypeSafe AI. We cite public docs. We do not sell access or imply a partnership.

Is this an OpenAI-compatible chat API?

No. System One is a Decisions API (https://api.typesafe.ai/v1/systemone). See the API quickstart.

Can Jev AI see images?

Not today. Input is text only: a string, a JSON object, or an array.

Does calibrated probability mean every answer is correct?

No. Official materials describe calibration over a population. Use confidence thresholds and human fallbacks for high-cost mistakes.

When should I still use a chat LLM?

Use a chat LLM when the artifact is language. Use Jev AI for the decision hop. See Jev AI vs LLM.

Where do I get an API key?

From the TypeSafe console at https://console.typesafe.ai. Keep keys on a server.

How do I estimate cost before integrating?

Use the pricing calculator. Public list price is $0.042/M input tokens. Output is free.

All 9 pages

1. HomeUnofficial handbook for Jev AI and TypeSafe AI: System One decision model API, noul/choice/score primitives, pricing, and when typed decisions beat a chat LLM.
2. TypeSafe AIWho TypeSafe AI is, founder Diogo Almeida, System One naming, published speed/cost claims, and official docs links.
3. API quickstartPOST /v1/systemone with noul, choice, and score in one call. Auth, errors, backoff, and why chat-completions SDKs will not work.
4. What is Jev AIJev AI is not a chat model. Learn noul, choice, score, confidence vs probability, and when not to use it.
5. PricingEstimate daily and monthly cost at $0.042 per million input tokens. Output tokens are free. Compare against a chat LLM.
6. PrimitivesField reference for noul, choice, and score plus a request builder that copies valid System One JSON.
7. Jev vs LLMParallel sampling vs autoregression, latency, calibrated probabilities, and using LLM write + Jev judge together.
8. ProjectsPublic GitHub projects for TypeSafe Jev AI: official SDKs, community clients, agent guardrails, routers, and demos.
9. Cases202 public Jev AI posts: original text, Chinese reference, metrics, and a link back to the post.