Skip to main content

Microsoft launches Decision-1 for fast decision scoring

17 NOVEMBER 2026·2 MIN READ·4 SOURCES

Microsoft announced Microsoft-Decision-1, a model for fast decision-scoring, on October 9, 2026, and says it is available through Microsoft Foundry and OpenRouter.

Key takeaways · 4

  • 01

    The model is in public preview in the Foundry Models catalog and is also available through OpenRouter.

  • 02

    Microsoft lists input pricing at $0.042 per million tokens and says output tokens are free.

  • 03

    Microsoft says its benchmark comparison covered 36 benchmarks and nearly 150,000 questions kept blind from training.

  • 04

    The catalog excludes using the model as the sole automated decision-maker for consequential personal decisions, including credit, employment and healthcare.

A model for structured decisions

Microsoft describes Decision-1 as a model for fast, single-pass decision scoring, created by post-training Qwen3.5-9B.[1] The company designed it for structured work such as routing, classification, prioritization, verification and workflow control.[1] For fixed answer options, Microsoft says it provides a calibrated probability score for each option.[1] It supports yes/no, multiple-choice and rating options, as well as rubric-based grading of AI responses and agent actions.[1] Microsoft’s Foundry blog presents decision models as separate building blocks that can complement generative and reasoning models.[2]

Availability and cost

Microsoft announced Decision-1 on October 9, 2026, and says it is available through Microsoft Foundry and OpenRouter.[1] Microsoft’s Tech Community blog describes it as in public preview in the Foundry Models catalog.[2] The listed input price is $0.042 per million tokens, with output tokens free, according to Microsoft.[1] That pricing and the fixed-option scoring approach give teams concrete details to assess when comparing it with other model choices; the evidence does not establish a total cost for a particular workload.[1]

Performance claims and measurement

Microsoft says Decision-1 had the highest accuracy in its comparison of 36 benchmarks covering nearly 150,000 questions kept blind from training.[1] The company reports that it was 2.5 times faster than H2O-Lightning-4B v1.1 and 35 times faster than GPT-6 Sol in its benchmarking.[1] Its launch comparison listed a median response time of 85 milliseconds for Decision-1 and 210 milliseconds for H2O-Lightning-4B, but the quoted figure used JevBench’s network-adjusted estimate rather than raw local execution time.[4] JevBench separately recorded a 459-millisecond native Azure Foundry median across a 124-request speed subset on October 10.[4]

Use cases and limits

Microsoft says Xbox Research used Decision-1 to sort more than 10,000 pieces of player feedback and reviews into researcher-defined themes.[1] Microsoft reports that this test was competitive in quality with GPT-6 Sol, ran over 14 times faster and cost 200 times less.[1] The company also says it tested safety on 5,250 requests across 11 benchmarks covering harmful content, jailbreak attempts and prompt injection.[1] Its catalog excludes using Decision-1 as the sole automated decision-maker for consequential personal decisions, including credit, employment and healthcare.[4] These results are company-reported findings, not evidence that the same performance will hold for every workload.

For teams building AI workflows, a dedicated scoring model may offer a distinct way to handle fixed-choice decisions alongside generative or reasoning models. The reported benchmark and pricing figures are useful starting points for evaluation, but teams should test their own workloads and respect the catalog’s stated limits on consequential decisions.

Why it matters
For teams

AI training for your team

kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.

How this developed

  1. 17 November 2026

    Microsoft launches Decision-1 for fast decision scoring

Sources

AI fluency, one session a day, built for your work.