# Microsoft-Decision-1 (slim)

> Microsoft-Decision-1 is a Microsoft model, post-trained from Qwen3.5-9B, that scores fixed answer options with a calibrated probability for each. It is in public preview on Microsoft Foundry and OpenRouter, at $0.042 per million input tokens.

- Full: https://www.anchorterminal.com/tools/microsoft-decision-1.md (~7,600 tokens) · this version ~2,130 tokens · JSON https://www.anchorterminal.com/tools/microsoft-decision-1.json · canonical https://www.anchorterminal.com/tools/microsoft-decision-1
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · Updated: 2026-10-10

**E · 39/100 · rank #921 of 954 · #14 in Decision models · not agent-ready · confidence medium**


Assessment: Microsoft publishes $0.042 per million input tokens, with output free, and the model is callable through Microsoft Foundry and OpenRouter. It is in public preview. No licence, model-specific retention statement, rate limit or deprecation policy was found, and the Foundry route needs an Azure deployment and an Entra token. Latency and accuracy figures are Microsoft's claims.

## Facts

- Kind: Model API · vendor: Microsoft · category: Decision models · legal entity: not named · provenance 41/100
- Local only (HTTP)
- Auth: OAuth or key · pricing: Pay per use · x402: no · licence: Not stated. The announcement and OpenRouter's model page name no licence, and no weights repository was found.
- Probe metrics: not measured yet (probes haven't run)
- Surface graded: Microsoft-Decision-1 on Microsoft Foundry (US and EU Datazone deployments) and on OpenRouter as microsoft/microsoft-decision-1. Read from Microsoft's post and OpenRouter's model page on 10 October 2026. No authenticated call was made.
- Base model: Post-trained from Qwen3.5-9B for single-pass decision scoring. Microsoft says future versions will be rebased on OpenAI and Microsoft AI (MAI) models.
- Question types: Microsoft says the model supports yes/no, multiple-choice and rating options, and rubric-based grading of AI responses and agent actions. Output is a probability for each option.
- Foundry request: POST to the deployment's /providers/microsoft/v1/systemone path, with the fields model, state and questions. Each question has type choice, instructions and criteria, which maps option labels to descriptions. The response carries an answers object whose choice must match a criteria key. Authentication is a Microsoft Entra bearer token for https://cognitiveservices.azure.com/.default.
- OpenRouter request: The Decisions method in OpenRouter's Alpha SDK (alpha.decisions.create with model, questions and state), authenticated with an OpenRouter API key. OpenRouter's documentation says chat completions SDKs will not work with this route.
- Context and output: OpenRouter lists 32,768 tokens of context. Microsoft's post and the Foundry sample do not state a context length or output limit.
- Accuracy claim (vendor): Highest accuracy in a 36-benchmark comparison of nearly 150,000 questions kept blind from training. No scores are published in the post.
- Perturbation claim (vendor): Decisions changed on 1.3 per cent of perturbations on average across eight perturbations of the same request. Paraphrased option descriptions and reordered or reversed options produced no changes.
- Calibration claim (vendor): A 90 per cent prediction should be right about nine times out of ten on representative cases. This is a stated goal, and the post gives no measured calibration result.
- Customer results (vendor): Xbox Research categorised more than 10,000 feedback items at quality competitive with GPT-5, 80 to 100 times faster (Tech Community post). The Command Line post gives quality competitive with GPT-6 Sol, over 14 times faster and 200 times less expensive. Copilot quality control was competitive with GPT5.6 Luna and 100 times faster. Microsoft Discovery adaptive replanning was 46 times more consistent than LLM-based scoring.
- Data handling: Microsoft's Foundry data page for Models sold by Azure: no prompts or completions stored in the model, no training on them, processing within the customer's geography unless a Global or DataZone type is used, and abuse monitoring stores prompts for human review unless the customer applies to change it. The page does not name Microsoft-Decision-1.
- Regions: US Datazone and EU Datazone deployment types, both $0.042 per million input tokens. No other region is listed in the reviewed pages.
- Preview status: Public preview, per the Tech Community post of 9 October 2026.
- Open weights: Not stated. No Microsoft model with this name was found on Hugging Face.
- Prices: Decision input $0.042 per 1M tokens; Decision output free per 1M tokens

- Scores: Reliability 30, Performance pending, Schema & documentation 43, Agent ergonomics 50, Security & auth 52, Payments & pricing 20, Task success pending, Maintenance & community 40, Transparency & trust 32 · total over the 7 assessed categories
- Why: Reliability, Azure status history has a Foundry Models component with dated incidents, earning 20 of 20 for a public status page with component history. · Schema & documentation, No machine-readable contract for the model was found (0 of 25). · Agent ergonomics, The response is one probability per option, and the caller sets the number of questions and options. · Security & auth, The Foundry sample uses a Microsoft Entra bearer token for a fixed resource scope, earning 20 of 30. Role assignments and token lifetimes we… · Payments & pricing, The price is published without a login at $0.042 per million input tokens, with output free, earning 20 of 20. No x402, MPP or L402 was foun… · Maintenance & community, The model was released on 9 October 2026, within 30 days, earning 30 of 30. The record holds one dated release, so there are no three releas… · Transparency & trust, No licence for the hosted model is stated in the announcement or on OpenRouter's page, and no Microsoft weights repository was found, so 0 o…
- Sources: 9, open questions: 9, both in the full twin
- Capabilities: inference.decision
- JSON: https://www.anchorterminal.com/api/v1/tools/microsoft-decision-1.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/microsoft-decision-1.svg` or a link to https://www.anchorterminal.com/tools/microsoft-decision-1 from a page on microsoft.com or one of its subdomains, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.

## Before you call it

1. Use OpenRouter's Decisions method, not an OpenAI chat-completions SDK. OpenRouter says chat completions SDKs will not work with this model
2. On Foundry, send the request to the deployment's /providers/microsoft/v1/systemone path with a Microsoft Entra token for https://cognitiveservices.azure.com/.default, not an API key
3. Take the deployment name from the Foundry quickstart before the first call. Microsoft says to confirm the route and authentication header, and the pages reviewed do not give the name
4. Keep each request within OpenRouter's 32,768-token context and send only fixed options, since the model is not intended for open-ended generation
5. Measure latency and calibration on your own labelled cases before relying on Microsoft's latency and 'nine times out of 10' statements

## Connect

```bash
curl -X POST "$AZURE_ENDPOINT/providers/microsoft/v1/systemone" \
  -H "Authorization: Bearer $ENTRA_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"model":"<deployment-name>","state":"The customer requests a refund for a duplicate charge.","questions":{"team":{"type":"choice","instructions":"Which team should handle this request?","criteria":{"billing":"Charges, invoices, refunds or subscription payments","technical":"Software errors or integration failures"}}}}'
```

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| OpenAI Decisions API | BB | 71.5 | inference.decision | https://www.anchorterminal.com/tools/openai-decisions-api.min.md |
| Drex 1.5 | B | 69.7 | inference.decision | https://www.anchorterminal.com/tools/nace-drex.min.md |
| Decider | B | 69.5 | inference.decision | https://www.anchorterminal.com/tools/decider.min.md |
| Laya | B | 69.2 | inference.decision | https://www.anchorterminal.com/tools/convai-laya.min.md |
| Kev | B | 67.4 | inference.decision | https://www.anchorterminal.com/tools/jaredpalmer-kev.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)