Felo API PlatformFelo API Platform
Back to models

Jev 1.14

TypeSafe's System One decision model: structured classifications, choices and yes/no judgements with calibrated confidence instead of long text.

Jev 1.14 is TypeSafe AI's decision model, the flagship of what the company calls System One models. It does not write prose — it returns a structured answer to a structured question: a category, a chosen option, a yes/no judgement, each paired with a calibrated confidence value. Outputs are meant to be thresholded in application code rather than read by a person.

Version 1.14.0, released October 14, 2026, adds three experimental capabilities: image input that can be combined with text, a Box data type for locating things such as a cup in a photo or a signature on a scanned page, and a context window expanded fourfold to 262k tokens. All three belong to TypeSafe's "danger zone" experimental program, are off by default, must be enabled through an API switch, and are explicitly not production-grade.

It fits high-volume, low-latency decisions: content classification, intent detection, routing, and the individual judgement calls inside an agent loop. TypeSafe reports classification up to 200× faster and 400× cheaper than comparable LLMs, at $0.042 per million input tokens with output free of charge. Non-English languages, including Chinese, still lag English, so validate on your own content before shipping.

Updated:

MODEL ID

jev-1.14
Context
262.144K
Max output
235.929K
Input
text / image
Output
decisions

Providers

1 providers

Compare API prices across providers. Select a provider to view all charges and conditions.

Provider · Pricing details
$0.042$0

Decisions Quick Start

Send state and narrow typed questions to /api/v1/decisions and read one answer per question: noul returns a probability from 0 to 1, choice the selected option with its distribution, and score a numeric score for ordered criteria. This endpoint is separate from the Chat Completions, Messages, and Responses surfaces.

curl https://openapi.felo.ai/api/v1/decisions \
  -H "Authorization: Bearer $FELO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "jev-1.14",
    "state": "Help! My payouts have been failing for 3 days.",
    "questions": {
      "is_urgent": {
        "type": "noul",
        "instructions": "Does this message convey urgency?",
        "criteria": {"true": "Time-sensitive", "false": "No urgency"}
      },
      "department": {
        "type": "choice",
        "instructions": "Which team should handle this?",
        "criteria": {
          "billing": "Payments, invoicing, refunds",
          "technical": "Bugs, outages, integrations",
          "sales": "Pricing, upgrades, new accounts"
        }
      },
      "frustration": {
        "type": "score",
        "instructions": "How frustrated is the customer?",
        "criteria": ["Calm", "Frustrated", "Very angry"]
      }
    }
  }'

Frequently asked questions

What is Jev 1.14?

Jev 1.14 is TypeSafe AI's decision model, the flagship of what the company calls System One models. It does not write prose — it returns a structured answer to a structured question: a category, a chosen option, a yes/no judgement, each paired with a calibrated confidence value. Outputs are meant to be thresholded in application code rather than read by a person. Version 1.14.0, released October 14, 2026, adds three experimental capabilities: image input that can be combined with text, a Box data type for locating things such as a cup in a photo or a signature on a scanned page, and a context window expanded fourfold to 262k tokens. All three belong to TypeSafe's "danger zone" experimental program, are off by default, must be enabled through an API switch, and are explicitly not production-grade. It fits high-volume, low-latency decisions: content classification, intent detection, routing, and the individual judgement calls inside an agent loop. TypeSafe reports classification up to 200× faster and 400× cheaper than comparable LLMs, at $0.042 per million input tokens with output free of charge. Non-English languages, including Chinese, still lag English, so validate on your own content before shipping.

How is usage billed?

Your bill depends on the provider you use, its prices and your usage. Longer inputs or other conditions may have different prices.