Sign in
Pending benchmarksTracking since Nov 17, 2026

Is Microsoft-Decision-1.1 state of the art?Pending benchmarks.Announced Nov 17, 2026; no independent results yet.

Microsoft's decision model, announced as generally available on November 17, 2026, for structured choices in software such as classification, prioritization, verification, workflow control, and routing. Official materials describe a mid-sized mixture-of-experts based on Gemma 4 26B A4B, with a custom inference engine, image and screenshot input, and calibrated confidence scores. Available in Microsoft Foundry and through OpenRouter and Vercel. Official Foundry Global Standard pricing is $0.042 per 1M input tokens, with output tokens free. No open weights.

Announced facts

The details so far

Lab
Microsoft
Announced
Nov 17, 2026
API
Live · microsoft-decision-1.1
Open weights
Closed · Not released
Parameters
Gemma 4 26B A4B mixture-of-experts (official GA post)
Focus
Structured software decisions: routing, classification, prioritization, verification, and workflow control, with calibrated confidence and the option to abstain (official GA post)
Architecture
Custom inference engine; official GA post says it is based on Gemma 4 26B A4B, a mid-sized mixture-of-experts
Input
Text, plus images such as photos and screenshots (official GA post)
Availability
Generally available in Microsoft Foundry, and through OpenRouter and Vercel (official GA post)
Price (Microsoft Foundry)
Input tokens: $0.042 per 1M (Global Standard), $0.046 per 1M (US Data Zone), $0.05 per 1M (EU and APAC Data Zone). Output tokens are free (official GA post)
Related
Successor to Microsoft-Decision-1 (not tracked here). Separate from Gemini Decision 1

Tags marked unsure are not stated on the lab's own pages.

Next steps

What would let us call it

Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.

  • Publicly announced with a source
  • Usable by the public (API or weights)
  • Listed on an independent board we track
  • Enough results for a rank in a category
  • Holds across two snapshots (crowns only)
Self-reported

What the lab claims

Self-reported. Shown for context; not used in any verdict.

Lab claim: self-reported official GA-post score of 68 on the Artificial Analysis Decision Index v2.1, six points above Microsoft-Decision-1, at the same cost and more than twice the speed of that earlier model.

Lab claim: self-reported official Ignite blog comparison that Decision 1.1 is 50 times faster and 79 times cheaper than GPT-Sol 6.1.

Receipts

Sources

LABSeven new world class models at scalable price-performance (Microsoft AI, Ignite)microsoft.ai/news/microsoft-ignite-2026/ · Nov 17, 2026LABMicrosoft-Decision-1.1 is now generally available (Microsoft commandline blog)commandline.microsoft.com/microsoft-decision-model-ga/ · Nov 17, 2026

Tracking continues until an independent board lists it.