Microsoft: Decision-1 API Reference

About

Microsoft-Decision-1 is a text decision model from Microsoft, served through Azure on Microsoft Foundry with a 32,768-token context. It reads text or structured JSON and returns typed decisions with calibrated probabilities for Noul, Choice, and Score questions, without generating free-form text or written rationales. The model supports classification, routing, ranking, and grading. The upstream accepts 2–255 choice options and 2–10 score levels.

1. Calling the API

Setup your API Key

To begin using Infron, create an account and generate your API Key.

export INFRON_KEY="YOUR_API_KEY"

Submit a request

Send JSON to POST https://llm.onerouter.pro/v1/decisions using your Infron API key. This is a runnable request example; route availability depends on your deployment.

curl --request POST "https://llm.onerouter.pro/v1/decisions" \
  --header "Authorization: Bearer $INFRON_KEY" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "microsoft/microsoft-decision-1",
  "state": "A customer reports being charged twice and asks for a refund today.",
  "questions": {
    "billing": {
      "type": "noul",
      "instructions": "Is this request about billing?"
    },
    "team": {
      "type": "choice",
      "instructions": "Which team should handle it?",
      "criteria": {
        "billing": "Payments and refunds",
        "technical": "Software support"
      }
    },
    "urgency": {
      "type": "score",
      "instructions": "How urgent is this request?",
      "criteria": [
        "Can wait",
        "Handle today",
        "Immediate emergency"
      ]
    }
  },
  "provider": {
    "only": [
      "azure"
    ],
    "allow_fallbacks": false
  },
  "usage": {
    "include": true
  }
}'

Response

The following single-answer response illustrates the schema. Its request ID, probabilities, and usage are placeholders, not a captured call.

{
  "id": "request_id_example",
  "request_id": "request_id_example",
  "model": "microsoft/microsoft-decision-1",
  "answers": {
    "billing": {
      "type": "noul",
      "noul": 0.97
    }
  },
  "usage": {
    "input_tokens": 100,
    "output_tokens": 1
  }
}

Routing and usage

provider and usage.include are Infron fields. They are not forwarded to the upstream provider. Set provider.only to ["azure"] and provider.service_tier to "standard" to select this route. usage.include=true includes cost information. Charges use the actual upstream usage.input_tokens; output tokens are free and their actual count is preserved.

Errors

Invalid parameter types or configured limits are rejected before the upstream request. Provider authentication, account, and context errors propagate through the existing gateway error handling. The service does not silently truncate the configured questions, options, or levels. Streaming and asynchronous task polling are unavailable.

2. Schema

Input

ParameterTypeRequiredDefaultRangeDescription
modelstringYes--microsoft/microsoft-decision-1Public model ID; upstream selector is the Azure deployment Decision-1.
statestring / object / arrayYes--See provider contractShared content for the questions.
questionsobjectYes--1–255 entriesNamed questions; names are retained in the response.
questions.<id>.typestringYes--noul, choice, scoreQuestion type.
questions.<id>.instructionsstring / object / arrayConditional--See Question typesWhat to evaluate; no default is injected.
questions.<id>.criteriaobject / arrayFor choice and score--See Question typesOptions or an ordered rubric.
providerobjectNo----Platform routing options.
usage.includebooleanNofalsetrue, falseInclude platform cost details.

Question types

Noul returns a yes probability; criteria is optional and takes true and false descriptions. Choice accepts 2–255 options; Score accepts 2–10 ordered levels, indexed from zero, and returns the probability-weighted average of the level indexes, so a result can fall between levels. Choice requires an option map; Score requires an array. A request can carry up to 255 questions. Instructions and criteria describe the task according to the upstream contract; Noul requires a meaningful question through instructions or criteria. The Microsoft Foundry catalog specifies a 32,768-token context.

Media inputs

Microsoft-Decision-1 is text-only. Image, audio, and video inputs are not accepted.

Output

ParameterTypeRangeDescription
id, request_idstring--Platform request ID.
modelstring--Public model ID.
provider, service_tierstring--Selected route.
answersobjectRequest question IDsTyped upstream answers.
answers.<id>.typestringnoul, choice, scoreAnswer type.
answers.<id>.noulnumber0–1Yes probability.
answers.<id>.choicestringChoice keysSelected option.
answers.<id>.scorenumber0 to levels minus 1Probability-weighted level.
answers.<id>.confidencenumber0–1Choice/Score confidence.
answers.<id>.probabilitiesobjectValues 0–1Probability distribution.
answers.<id>.legendobject--Score level descriptions.
usage.input_tokens, usage.output_tokensintegerNon-negativeActual upstream usage.
costnumberUSDCharge details when requested.
cost_detailsobject--Billing breakdown when requested.

3. Detailed Pricing

ModelBilling unitInput priceOutput price
Microsoft-Decision-1Per 1 million tokens$0.042Free

This is the undiscounted base price; no promotional discount is configured. Charges use the actual upstream usage.input_tokens; output tokens are free and their actual count is preserved. Official pricing.