# TypeSafe AI's Jev: Typed AI Decisions for Code, Explained

> TypeSafe AI's Jev answers your code with typed decisions and a confidence score, in milliseconds, for $42 per billion tokens. What it is, the code, and the fine print.

By Salaheddine Elfatimi, Full Stack Developer in Marrakech. Published 27 September 2026. Updated 27 September 2026.
Canonical: https://salaheddine-elfatimi.com/blog/typesafe-ai-jev-system-one-model

## Key takeaways

- TypeSafe AI's Jev is a 'System One Model': it returns typed decisions with a confidence score instead of text.
- Questions come in three types: Choice (pick an option), Score (a scale) and Noul (a 0–1 yes/no).
- TypeSafe quotes 70–500 ms per call and $42 per billion input tokens, with output free.
- By TypeSafe's own account, its 193.6× faster / 444.6× cheaper figures are on the high end, and the pricing can't yet be proven unsubsidized.
- Jev is invite-only for now, with a waitlist on typesafe.ai.

**TypeSafe AI** is a San Francisco AI lab that released **Jev**, which it calls the first public "System One Model". Instead of writing text like a chatbot, Jev answers questions your code asks with a **typed decision and a confidence score**, in 70 to 500 milliseconds, for $42 per billion input tokens. It's built for the small decisions inside software: route this message, is it urgent, is that AI answer correct.

![The official quickstart example: one call returns a team, a frustration score and an urgency value.](https://salaheddine-elfatimi.com/uploads/2026/09/typesafe-ai-jev-cover-3f8abd14.webp)

I read the [launch post](https://typesafe.ai/blog/introducing-system-one-models-and-jev), the [docs](https://docs.typesafe.ai/introduction/quickstart) and the home page. Jev is invite-only for now, so I haven't run it myself yet: everything below is what TypeSafe publishes, with their own caveats included.

## What is TypeSafe AI's System One Model?

The name comes from the idea of fast, intuitive thinking ("System One") versus slow, deliberate reasoning. In the [launch post](https://typesafe.ai/blog/introducing-system-one-models-and-jev), founder Diogo Almeida, who worked at OpenAI on the methods behind ChatGPT, describes it as "a new class of frontier models built to make fast, structured decisions that software can use directly."

His one-line summary: "Think of Jev as a frontier-intelligence function call: unstructured state in, typed probabilistic decisions out."

|  | Chat models (LLMs) | System One + Jev |
| --- | --- | --- |
| Output | Free text, to parse and validate | Typed values defined in advance |
| Confidence | Often overconfident, even when asked | A confidence with every answer |
| Speed (their figures) | 3 to 329 seconds end to end | 70 to 500 ms end to end |
| Input price | $0.20 to $10 per million tokens | $0.042 per million tokens |
| Output price | About 5× the input price | Free ("too cheap to meter") |
| Trained with | RLHF / RLVR | RLCD: Reinforcement Learning for Calibrated Decisions |

## How does Jev work in code?

You send a piece of text (the "state") and a set of questions. Each question has a type that fixes what the answer can be:

- **Choice**: one option from a list you define ("billing", "technical", "sales"), with probabilities and a confidence.
- **Score**: a position on a scale you describe, with probabilities and a confidence.
- **Noul**: a yes-or-no statement answered as a number from 0 to 1 ("the message conveys urgency").

This is the Python example from the quickstart (install with `pip install typesafe-sdk`, key from the TypeSafe console):

```
from typesafe_sdk import Choice, Noul, Score, TypeSafeClient

client = TypeSafeClient()

ticket = "Hi, I've been trying to connect my Stripe account for 3 days and the integration keeps failing. I'm losing sales. Please help ASAP."

response = client.system_one(
    state=ticket,
    questions={
        "department": Choice(
            instructions="Which team should handle this",
            criteria={
                "billing": "Payment or subscription issues",
                "technical": "Bugs or integration problems",
                "sales": "Pricing or account questions",
            },
        ),
        "frustration": Score(
            instructions="How frustrated the customer appears",
            criteria=[
                "Calm, just stating facts",
                "Frustrated but civil",
                "Very angry, strong language",
            ],
        ),
        "is_urgent": Noul(
            instructions="The message conveys urgency or time-sensitivity",
        ),
    },
)

print(response.answers["department"].choice)  # "technical"
print(response.answers["frustration"].score)  # 1.0
print(response.answers["is_urgent"].noul)     # 1.0
```

![The quick start also has a Playground: paste a message, add a question, see the typed answer.](https://salaheddine-elfatimi.com/uploads/2026/09/typesafe-ai-jev-docs-01b6d3c1.webp)

## Why the confidence score matters

A chat model that's right 95% of the time is hard to automate with if it doesn't tell you when it's in the other 5%. TypeSafe's pitch is that Jev's confidence is calibrated: "higher confidence means higher accuracy". So your code, not the model, draws the line: above your threshold it acts on its own, below it a person takes a look.

![You pick the threshold: confident answers are handled automatically, the rest go to a person.](https://salaheddine-elfatimi.com/uploads/2026/09/typesafe-ai-jev-flow-bd31ad38.webp)

## The numbers, and TypeSafe's own fine print

The home page headline is "193.6x Faster, 444.6x Cheaper", with a side-by-side run: $0.000081 and 0.114 s for Jev against $0.013880 and 8.566 s for an LLM. It also says Jev's input price is "238x Lower" than Claude Fable 5.1's (Fable is $10 per million input tokens; Jev $0.042).

To TypeSafe's credit, the [launch post](https://typesafe.ai/blog/introducing-system-one-models-and-jev) spells out the limits of these claims:

- The 193.6× and 444.6× figures are, in their words, "on the higher end of real world gains".
- On price: "We can't prove it isn't subsidized".
- The workflow evals were made by their own team, "so some bias could exist", and speed was measured from the US West Coast, where the service runs.
- "Zero hallucinations" means no type errors: the answer is always one of the allowed values. Their own FAQ still asks "Can Jev still get things wrong?", which is why the confidence score matters.

![The launch post, 15 September 2026, with the benchmarks and the caveats.](https://salaheddine-elfatimi.com/uploads/2026/09/typesafe-ai-jev-launch-f9970961.webp)

## Where it could help a business

The launch post lists "smart if-statements" inside workflows, map-reducing over big data, real-time features, and checking other AI output ("score, judge, verify, guardrail"). Translated to the projects I build:

- **Routing incoming messages**: a contact form or WhatsApp message goes to sales, support or billing, with urgent ones flagged first.
- **Checking a chatbot's answer** before a customer sees it: is it on topic, does it promise something it shouldn't?
- **Tagging a backlog**: thousands of tickets or reviews sorted by topic and mood for the price of a coffee.

For anything that needs writing (a reply, a summary, code), you still need a chat model: Jev "gives up string generation" by design. The interesting setup is both together, with Jev making the fast yes/no and routing decisions around the slower model.

## How can you try TypeSafe AI's Jev?

Jev launched in early access, and the home page now says "TypeSafe is invite-only again", with a waitlist. Existing users can log in to the console and the Playground. The docs, including a guide for coding agents, are public at [docs.typesafe.ai](https://docs.typesafe.ai/).

![The TypeSafe AI home page, with the waitlist box.](https://salaheddine-elfatimi.com/uploads/2026/09/typesafe-ai-jev-home-262069a8.webp)

I'll test it on a real routing workflow once I get access and write up the results. Meanwhile, if you want AI inside your platform, like a CRM that sorts its own messages or a support inbox that knows what's urgent, [tell me about your project](https://salaheddine-elfatimi.com/#contact). More AI notes: [September's developer news](https://salaheddine-elfatimi.com/blog/nextjs-security-update-september-2026) and [what WebMCP is](https://salaheddine-elfatimi.com/blog/what-is-webmcp).

## Frequently asked questions

### What is TypeSafe AI?

A San Francisco AI lab, founded by Diogo Almeida, that builds 'System One Models': AI models that give software typed decisions with confidence scores instead of generated text. Its first public model is Jev, launched on 15 September 2026.

### Is Jev a chatbot or a smaller LLM?

No. Jev doesn't generate text. You send it text and questions with predefined answer types (Choice, Score, Noul) and it returns the answers with probabilities and a confidence.

### How much does Jev cost?

TypeSafe lists $0.042 per million input tokens ($42 per billion) with output free. The company says it can't yet prove the price isn't subsidized and expects it to go down.

### Can Jev be wrong?

Yes. 'Zero hallucinations' refers to type safety: the answer is always one of the allowed values. The answer itself can still be wrong, which is what the confidence score is for.