Updated

Microsoft-Decision-1

Microsoft introduced Microsoft-Decision-1, a decision-scoring model for structured tasks such as routing, classification, prioritization, and verification. It returns calibrated scores for fixed choices and is available through Microsoft Foundry and OpenRouter.

By No Code MBA · Facts checked

Provider
Microsoft
Announcement date
Status at last check
Generally available
Capabilities
Text

What changed

Microsoft-Decision-1 is a decision-scoring model post-trained from Qwen3.5-9B. It returns a calibrated probability for each option in a fixed set (yes/no, multiple choice, ratings) and can grade AI responses and agent actions against a rubric in a single pass. Microsoft reports the highest accuracy across a 36-benchmark comparison of about 150,000 held-out questions, and says it is about 2.5x faster than H2O-Lightning-4B v1.1 and about 35x faster than GPT-6 Sol at median latency.

Access and availability

Deploy it from the Microsoft Foundry model catalog as Microsoft-Decision-1, or call it on OpenRouter as microsoft/microsoft-decision-1. Requests use a structured API call listing the choices to score.

Official access documentation

Pricing

$0.042 per million input tokens; output tokens are free

Official pricing

What you can build

Evaluate it on a representative routing or classification task with a fixed set of choices, and measure decision quality and latency against the current approach.

Limitations

It only scores fixed answer options and does not generate free-form text. Microsoft advises checking calibration on representative cases, plans to rebase the model on other models soon, and did not state license or context length.

Sources and release history