Jev AI, explained. DecisionNode, measured against it.

Jev turns application state into typed decisions. What it does, what it costs, and a measured alternative that reads images.

Try DecisionNode Compare the two

Accuracy, measured October 2026

0 to 1, higher is better. Full axis, nothing cut off.

Mean of 15 text suites

DecisionNode MoE 0.816
TypeSafe Jev 1.13 0.803

Prompt injection

DecisionNode MoE 0.764
TypeSafe Jev 1.13 0.600

Photos (image suite)

DecisionNode MoE 0.918
TypeSafe Jev 1.13 not supported, text only

Jev leads on knowledge questions: 0.820 against 0.698. Preliminary results, October 2026. Every suite and the method

What Jev does

Jev is TypeSafe AI's hosted decision model. You send it state, such as a support ticket or a JSON record, together with typed questions. It answers each question with probabilities instead of text, so your code can branch on the result. TypeSafe calls it a System One model.

The full Jev explainer

Choice choice: One option from up to 255, with a probability for every option and a confidence from 0 to 1. Used for: Routing, intent, classification.
Score score: A place on 2 to 10 ordered levels: the expected level, the probabilities and a confidence. Used for: Urgency, risk, lead quality, review priority.
Yes/No noul: One probability from 0 to 1 that the answer is yes. Your code picks the cut-off. Used for: Gates before an action, policy checks.

Price: $0.042 per 1M input tokens, output free
Input: Text and JSON only
Context: 64k tokens a request
Rate limit: 80 requests a second per key

Source: TypeSafe's documentation, checked October 12, 2026.

Where DecisionNode goes further

Both answer typed questions with probabilities. DecisionNode reads more kinds of input, answers more kinds of question, and is cheaper per token on its Flash tier.

TypeSafe Jev 1.13 and DecisionNode compared
Feature	TypeSafe Jev 1.13	DecisionNode
Reads	Text and JSON	Text, JSON, images (up to 16 a request) and video frames
Answer types	Choice, score, yes/no	Choice, score, yes/no, number, point and box
Accuracy, mean of 15 text suites	0.803	0.816 (MoE), 0.784 (Flash)
Prompt injection suite	0.600	0.764 (MoE)
Price per 1M input tokens	$0.042	$0.042 (MoE), $0.035 (Flash)
Batch jobs	Not in Jev's documented API	30% off on MoE, 50% off on Flash
Streaming sessions	Not in Jev's documented API	WebSocket sessions for steady streams of decisions
Free tier	None listed by TypeSafe	50 Flash and 20 MoE requests a day

Jev: TypeSafe's documentation. DecisionNode: decisionnode.com. Both checked October 12, 2026. Accuracy: preliminary results, october 2026.

Where Jev is ahead

Knowledge questions (MMLU-Pro): 0.820 for Jev against 0.698 for DecisionNode MoE.
New task families: 0.823 against 0.797.
Official SDKs for Python and JavaScript. DecisionNode is called over plain HTTPS.
Training: TypeSafe does not train on customer requests. DecisionNode does unless you opt out.

The full comparison, suite by suite

State in, typed answers out

DecisionNode requests exactly as you would send them, and what each question returns. The same text requests also run on Jev.

Support triage

Route a ticket, rate its urgency and decide whether a person should take it, before any reply is written.

POST /v1/decide

{
  "model": "decisionnode-1.0",
  "state": "We were charged twice for the October invoice (INV-20931). The second charge pushed our card over its limit and our team seats were suspended this morning. Please refund the duplicate and restore access.",
  "questions": {
    "route": {
      "type": "choice",
      "instructions": "Which team should handle this ticket?",
      "criteria": {
        "billing": "Charges, invoices, refunds and plans",
        "bug": "Something in the product is broken",
        "account": "Login, seats and permissions",
        "feature_request": "Asks for something the product does not do"
      }
    },
    "urgency": {
      "type": "score",
      "instructions": "How urgent is this ticket?",
      "criteria": [
        "routine: no deadline",
        "soon: within a few days",
        "today: blocks work today",
        "critical: outage or data loss"
      ]
    },
    "refund": {
      "type": "truth",
      "instructions": "Is the customer asking for money back?",
      "criteria": {
        "true": "Asks for a refund or a credit",
        "false": "Any other request"
      }
    },
    "escalate": {
      "type": "truth",
      "instructions": "Should a person handle this now instead of an automated reply?"
    }
  }
}

What comes back

route: choice, one of 4 options with a probability for each
urgency: score on 4 levels, with probabilities and a confidence
refund: yes/no, one probability from 0 to 1
escalate: yes/no, one probability from 0 to 1

Run this request on DecisionNode

Agent tool-call guardrail

Check a tool call an agent is about to make against what the user asked for, and block the destructive ones.

POST /v1/decide

{
  "model": "decisionnode-1.0",
  "state": {
    "user_request": "Clean up the test accounts we created for the demo last week.",
    "proposed_tool_call": {
      "tool": "sql.execute",
      "database": "production",
      "query": "DELETE FROM accounts WHERE created_at < '2026-10-01';"
    }
  },
  "questions": {
    "verdict": {
      "type": "choice",
      "instructions": "Should the agent run `proposed_tool_call`?",
      "criteria": {
        "allow": "Safe, and does what the user asked",
        "ask_human": "Plausible, but needs a person to confirm",
        "block": "Unsafe, or does more than the user asked"
      }
    },
    "risk": {
      "type": "score",
      "instructions": "How much damage could `proposed_tool_call` do?",
      "criteria": [
        "read only",
        "reversible change",
        "hard to reverse",
        "destructive"
      ]
    },
    "in_scope": {
      "type": "truth",
      "instructions": "Does `proposed_tool_call` do only what `user_request` asks for?"
    }
  }
}

What comes back

verdict: choice, one of 3 options with a probability for each
risk: score on 4 levels, with probabilities and a confidence
in_scope: yes/no, one probability from 0 to 1

Run this request on DecisionNode

All five use cases

What a million decisions cost

At 300 input tokens a decision and list prices. Output is free on both.

TypeSafe Jev 1.13: $12.60
DecisionNode MoE: $12.60
DecisionNode Flash: $10.50
DecisionNode Flash, batch: $5.25

Each model counts tokens with its own tokenizer, so the same request can count differently. Both APIs report input tokens on every answer: compare those for your own requests. Jev pricing in detail

Jev questions

What is Jev?

Jev is a hosted decision model from TypeSafe AI. It takes text or JSON state plus typed questions and returns a choice, a score or a yes/no probability for each question, without generating text. The current version is Jev 1.13. What is Jev, in full

How much does Jev cost?

TypeSafe lists $0.042 per million input tokens ($42 per billion), and output tokens are free. Its documentation lists no free tier as of October 2026. Jev pricing in detail

Can Jev read images?

No. TypeSafe documents Jev as text only: images, audio and video have to be turned into text first. DecisionNode reads images and video frames directly, and scores 0.918 (MoE) on the Photos benchmark suite.

Is DecisionNode a drop-in replacement for Jev?

No. Both take state and a map of typed questions, but the endpoint, the keys, the model names and the name of the yes/no type differ (truth on DecisionNode, noul on Jev), so moving an integration means a small change to the client code.

Who publishes this site?

DecisionNode, made by Bynn Intelligence, Inc. The site is not affiliated with TypeSafe AI. Facts about Jev come from TypeSafe's public documentation and are dated; benchmark figures come from decisionnode.com/benchmarks.

Run your own decisions on DecisionNode

Send text, JSON or images with typed questions and get probabilities back. Free tier: 50 Flash and 20 MoE requests a day.

Try DecisionNode