Decide in milliseconds.
Act on the probability.
Microsoft-Decision-1 is Microsoft's decision model. It never writes text. You send a situation and typed questions, and it returns typed answers with calibrated probabilities: noul for yes/no, choice for a label, score for a rubric. $0.042 per million input tokens, and output tokens are free.
Also available: Jev by TypeSafe and Decider 27B by Perplexity, same contract.
83.5%
average accuracy, 36 benchmarks
85 ms
median latency
92.2
calibration, 100 = perfect
$0.042
per million input tokens, output free
Accuracy, latency and calibration as reported by Microsoft on 9 Oct 2026, measured in Microsoft Foundry. Not our measurement.
Category, urgency and escalate-to-human before any model writes a reply.
Free starter credits cover thousands of decisions. Or copy the code below and run it with your own key.
Answers arrive here as probabilities, one card per question. No text to parse, nothing invented outside the options you gave.
Jev decides in ~100 ms, then a chat model drafts only the tickets a bot may answer.
What to do with the numbers
- urgent > 0.7 and category = billing → billing on-call queue.
- escalate < 0.4 → let companyfabric/cheap draft the reply; otherwise a human answers.
- escalate between 0.4 and 0.6 is unsure: fail closed, human.
- Log churn_risk with the ticket; tune thresholds from the log, not from guesses.
This exact request, in code
curl https://api.companyfabric.com/v1/decisions \
-H "Authorization: Bearer $COMPANYFABRIC_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "microsoft/microsoft-decision-1",
"state": "Second time this month my card was charged twice. I want a refund today or I'\''m cancelling and disputing with my bank.",
"questions": {
"category": {
"type": "choice",
"instructions": "What is this ticket about?",
"criteria": {
"billing": "Charges, invoices, refunds, double charges",
"technical": "Bugs, errors, outages, something not working",
"account": "Login, access, password, settings",
"other": null
}
},
"urgent": {
"type": "noul",
"instructions": "Does this need a human reply within the hour?",
"criteria": {
"true": "Money lost, service blocked, or a stated deadline",
"false": "Can wait for the next business day"
}
},
"escalate": {
"type": "noul",
"instructions": "Should a human handle this instead of an automated reply?"
},
"churn_risk": {
"type": "score",
"instructions": "How likely is this customer to leave?",
"criteria": [
"No signal",
"Frustrated",
"Threatens to cancel or dispute"
]
}
}
}'Needs a CompanyFabric key, or add your own TypeSafe key in the BYOK vault and pay them directly at 0% fee. Cost comes back in usage.cost_usd.
One call.
Many answers.
Microsoft serves Microsoft-Decision-1 through Microsoft Foundry. CompanyFabric runs it on the same endpoint as Jev and Decider 27B, so the request you write once runs on any of them.
Send a state
A ticket, a brief, a transcript, a JSON record. It is sent once, whatever you ask about it.
Ask typed questions
noul is a yes/no probability. choice picks one label from criteria you describe. score places on a rubric you define, from zero. Ask several in one call.
Threshold the numbers
Answers are probabilities and distributions, never prose. Act above a threshold, hand to a human below it, and treat a noul near 0.5 as unsure.
Microsoft's decision model, next to every other model.
One key, one balance
The same API key and prepaid balance as every chat model in the catalog. No Azure subscription, Foundry resource or deployment to set up.
Same contract as the others
Every decision model here answers POST /v1/decisions with the same request and the same answer shapes. Switch between them by changing one field, "model".
Compare on your own data
The playground runs each starter, or your own state, on Microsoft-Decision-1, Jev and Decider 27B. Benchmarks are someone else's data; yours is the one that counts.
A cap on every key
Set a spend cap per API key, so one service or agent can never run past the budget you gave it.
One field picks the model.
The support triage starter, sent to Microsoft-Decision-1. Change "model" to "typesafe/jev-latest" and the same request runs on Jev.
curl https://api.companyfabric.com/v1/decisions \
-H "Authorization: Bearer $COMPANYFABRIC_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "microsoft/microsoft-decision-1",
"state": "Second time this month my card was charged twice. I want a refund today or I'\''m cancelling and disputing with my bank.",
"questions": {
"category": {
"type": "choice",
"instructions": "What is this ticket about?",
"criteria": {
"billing": "Charges, invoices, refunds, double charges",
"technical": "Bugs, errors, outages, something not working",
"account": "Login, access, password, settings",
"other": null
}
},
"urgent": {
"type": "noul",
"instructions": "Does this need a human reply within the hour?",
"criteria": {
"true": "Money lost, service blocked, or a stated deadline",
"false": "Can wait for the next business day"
}
},
"escalate": {
"type": "noul",
"instructions": "Should a human handle this instead of an automated reply?"
},
"churn_risk": {
"type": "score",
"instructions": "How likely is this customer to leave?",
"criteria": [
"No signal",
"Frustrated",
"Threatens to cancel or dispute"
]
}
}
}'- Model id
- microsoft/microsoft-decision-1
- Provider
- Microsoft
- Price
- $0.04 / M input · $0 / M output
- Endpoint
- POST /v1/decisions
- Question types
- noul · choice · score
- State
- text or JSON
Metered on input, billed to your prepaid balance; each response carries its exact cost in usage.cost_usd.
Questions about Microsoft-Decision-1.
What is Microsoft-Decision-1?
Microsoft-Decision-1 is Microsoft's decision model, launched on 9 October 2026 in Microsoft Foundry. It does not write text: you send a state and typed questions, and it returns one typed answer per question with a calibrated probability attached. noul is a yes/no probability, choice picks one label from criteria you describe, and score places the state on a rubric you define. On CompanyFabric you call it as "microsoft/microsoft-decision-1" on POST /v1/decisions.
How does Microsoft-Decision-1 compare with Jev and other decision models?
Microsoft reports it first on accuracy (83.5% average across 36 benchmarks kept blind from training) and the fastest it measured, at 85 ms median latency, with a calibration score of 92.2 out of 100. Those are Microsoft's numbers, not ours. All three decision models on CompanyFabric take the same /v1/decisions request and answer the same question types, so the playground runs any starter on each and you can compare them on your own data by changing "model".
Do I need an Azure or Microsoft Foundry account?
No. CompanyFabric serves Microsoft-Decision-1 from its own Foundry deployment, so your CompanyFabric API key and prepaid balance work for it the same way they do for every chat model. There is no Azure subscription, resource or deployment to set up.
How much does Microsoft-Decision-1 cost?
$0.042 per million input tokens, and output tokens are free. You are metered on what you send (state and questions), and the exact figure for each call comes back in usage.cost_usd. It is billed to the same prepaid balance as chat, and the welcome credit covers it.
Can I use Microsoft-Decision-1 commercially?
Commercial use per Microsoft Foundry terms; hosted API only, no weights.
Put a decision
in front of the spend.
POST /v1/decisions with "model": "microsoft/microsoft-decision-1" from your code, or let your agent call the decide tool on our MCP server before every expensive step.