TypeSafe AI: Jev

Overview

TypeSafe AI's first System One model, offered by TypeSafe in early access through its API. You send a state (text or JSON) and typed questions (choice, score, or a yes or no noul) and get typed answers with probabilities and a confidence, not free text. TypeSafe names classifying, routing, scoring, extracting, branching and checking LLM output as its uses. API only: no weights are published. TypeSafe's docs give its context as 64k tokens per request and 32k per question.

Provider

TypeSafe AI

Modality

Structured decisions

Input

text

Output

decisions

Weights

Closed

Distillable

NoLicence or terms, checked October 15, 2026.

Tags

system-one, structured-outputs, classification, routing, scoring, guardrails

Docs

Provider documentation

Data policy

Prompts go to

TypeSafe AI, through its API. Its Privacy Policy says its Services are hosted in the United States.

Retention

TypeSafe's Privacy Policy says it will not train or fine tune models on your prompts or other input, and keeps personal data as long as reasonably necessary to provide its Services. TypeSafe offers zero data retention to its enterprise customers only.

Zero data retention

Not offered

Region

United States: TypeSafe AI's Privacy Policy says its Services are hosted in the U.S.

Policy

TypeSafe AI data policy

Summarised from the provider's published terms. The provider's own policy is what applies.

Where you can run it

How Jev can be deployed, bought and reached through Safeguard.

Deployment

Public cloud

Procurement

Not published yet

Inference regions

🇺🇸United StatesCloud provider's availability page, checked October 15, 2026.

Closed weights stay with the provider, so this model runs in the provider's cloud, never on-prem or air-gapped.

Pricing

Notes

TypeSafe lists $0.042 per 1M input tokens ($42 per billion) and output tokens as free, on its models page, read 2026-10-15.

Source

TypeSafe AI pricing , last verified October 15, 2026

No markup on model usage: you pay the provider's list price.

API

Jev answers typed questions about a state you send: JSON in, JSON out, through the gateway's System One endpoint (not the chat API). Your API key and access to this model are set up for your workspace when you choose Use with Safeguard.

import requests

response = requests.post(
    "https://api.safeguard.sh/gpt/v1/systemone",
    headers={"Authorization": "Bearer SAFEGUARD_API_KEY"},
    json={
        "model": "typesafe/jev",
        "state": "…",
        "questions": {
            "is_urgent": {"type": "noul", "instructions": "Does this convey urgency?"},
        },
    },
)
print(response.json()["answers"])

Endpoint https://api.safeguard.sh/gpt/v1/systemone, model typesafe/jev.

Early access: Jev is not callable through the API yet. This request works once Safeguard opens it and switches the gateway on for your workspace.

TypeSafe AI offers Jev in early access. Request it for your workspace in the Console: Safeguard staff review each request and turn the model on for approved workspaces. Request access