---
title: "Microsoft releases Decision-1, a small model that picks between fixed options instead of writing text"
url: https://anawords.com/story/microsoft-decision-1-decision-scoring-model-foundry
published: 2026-10-12T12:47:35+05:30
updated: 2026-10-12T12:47:35+05:30
section: Models & Launches
author: Mayank Sahu
checked: sources checked by AI, approved by editor
version: 1
---

# Microsoft releases Decision-1, a small model that picks between fixed options instead of writing text

Built on Qwen3.5-9B, it returns a probability for each choice for tasks like routing and grading. Microsoft lists it at $0.042 per million input tokens, with output free.

**In short:** Microsoft released Microsoft-Decision-1 on 9 October 2026. The model does not write free text: given a fixed set of options, it returns a calibrated probability for each. It is post-trained from Qwen3.5-9B and costs $0.042 per million input tokens in Microsoft Foundry.

## What launched

Microsoft released Microsoft-Decision-1 on 9 October 2026.[^1] It is a decision-scoring model: you give it a question and a fixed set of options, and it returns a calibrated probability for each option in a single structured API call, instead of generating open-ended text.[^1][^2]

Microsoft says it is meant for routing, classification, prioritisation, verification and workflow control, including grading AI responses and agent actions against a rubric.[^1] It supports yes/no, multiple-choice and rating formats.[^1][^2]

## How it is built

The model is post-trained from Alibaba’s open-weight Qwen3.5-9B for single-pass scoring.[^1][^2] Microsoft says it will soon rebase the model on other models, including Microsoft AI and OpenAI models.[^1]

## Price and availability

Microsoft lists $0.042 per million input tokens, with output tokens free.[^1][^2] It is available in Microsoft Foundry, and Microsoft’s post also lists OpenRouter.[^1] TestingCatalog described OpenRouter support as coming soon at the time it reported.[^2]

## Microsoft’s claims

Microsoft says the model had the highest accuracy in a 36-benchmark comparison of nearly 150,000 questions held out from training, and that its median response time is about 35 times faster than GPT-6 Sol.[^1] It also says decisions changed on 1.3% of reworded inputs on average.[^1] These are Microsoft’s own results and have not been checked independently.

Microsoft also cites internal teams. For example, it says Xbox Research found quality competitive with GPT-6 Sol at over 14 times the speed and 200 times lower cost.[^1]

## Who made it

RuntimeWire reports that the post is by Achint Srivastava, a Microsoft vice president who earlier co-founded the AI evaluation start-up Pi Labs.[^3]

## What is not stated

Microsoft’s post does not state a licence for the model’s weights.[^1]

## Fact ledger

- [Confirmed] Microsoft released Microsoft-Decision-1 on 9 October 2026. (source 1,2,3)
- [Confirmed] It returns a calibrated probability for each of a fixed set of options. (source 1,2)
- [Confirmed] It is post-trained from Qwen3.5-9B. (source 1,2)
- [Confirmed] Price: $0.042 per million input tokens; output tokens are free. (source 1,2,3)
- [Company claim] Highest accuracy across 36 benchmarks; about 35x faster than GPT-6 Sol. (source 1)
- [Not known] The licence for the model weights.

## Sources

1. [Microsoft: Microsoft-Decision-1 in Foundry](https://commandline.microsoft.com/microsoft-decision-1-model-foundry/) — official
2. [TestingCatalog: Microsoft launches Decision-1 in Foundry](https://www.testingcatalog.com/microsoft-launches-decision-1-model-in-foundry/)
3. [RuntimeWire: Microsoft puts a 9B decision model in Foundry](https://runtimewire.com/article/microsoft-decision-1-achint-srivastava-agent-workflows)