What is Microsoft-Decision-1?

Microsoft-Decision-1 is the decision-scoring model Microsoft unveiled on 9 October 2026 — its entry into the category Jev opened. Instead of writing text, it takes context plus a fixed set of answers and returns a structured result with a probability per option: a yes/no, a pick from a set, or a score. It's the biggest vendor yet to ship a decision model. Here's what it is, how it's built, where you can run it, and how it compares to Jev.

A decision model doesn't generate prose — it decides. You hand it some state (a message, a ticket, a proposed action) plus a finite answer space, and it returns which answer, with a calibrated probability, in a single pass your software can act on. Microsoft-Decision-1 is Microsoft's take on that idea; Jev is TypeSafe AI's purpose-built, hosted one. Same call shape, different build.

What Microsoft-Decision-1 is

Microsoft describes it as a decision-scoring model for routing, classification, prioritization, verification and workflow control. Rather than writing text, it emits structured output with a probability for each option — the reported task shapes are yes/no, multiple-choice and scoring. It was further-trained from Alibaba's open-source Qwen3.5-9B, and Microsoft says later versions will be based on its own MAI model and on OpenAI models. It's available first through Microsoft Foundry, with OpenRouter listed as coming soon.

The quick facts

• Vendor: Microsoft — announced 9 October 2026, the largest vendor in the decision-model category so far
• Backbone: post-trained from the open-source Qwen3.5-9B (future versions on MAI + OpenAI models)
• Output: a probability per option — yes/no, pick-one, or a score — not generated text
• Where: Microsoft Foundry first, OpenRouter coming soon; low per-token input pricing, output free (the category norm)
• Benchmarks: Microsoft's speed and accuracy numbers are vendor-measured on its own tasks, not an independent lab

How to use it

Because it lives in Microsoft Foundry first, Microsoft-Decision-1 fits most naturally if your stack is already on Azure / Foundry / Copilot — you call it where your other infrastructure already is. You send context plus a fixed answer space and read back a probability per option, then branch on it in code. OpenRouter access is listed as coming, which would let you reach it through a gateway alongside other models.

Read the speed claims honestly

Microsoft's headline numbers come from internal teams, not a third-party lab: its Xbox team reported classifying 10,000+ pieces of gamer feedback at roughly a two-hundredth of the cost and 14x+ the speed of GPT-6 Sol at similar quality, and a Copilot team reported ~100x the speed of GPT-5.6 Luna. Those are genuine but vendor-measured against specific large models on Microsoft's own tasks — the same caveat that applies to any vendor's figures, Jev's included. The only benchmark that settles it is your own decision, on your own data.

Microsoft-Decision-1 vs Jev

Both answer in the same shape — context plus a finite answer space in, a calibrated probability out — so the choice is about build and availability, not capability. Microsoft-Decision-1 post-trains an existing open model (Qwen3.5-9B) and lives in Foundry; Jev is a purpose-built System One model, RLCD-calibrated with pinned versions, generally available with a self-serve key and a free browser playground. We break the head-to-head down field by field on the dedicated comparison.

Try a decision now

Both models take the same thing: state plus typed questions, a calibrated decision back. The fastest way to feel the category is to run one — Jev's playground does it in the browser with no signup, and a jv_live_ key points your code at the hosted endpoint:

// A typed, calibrated decision — the same context + fixed-answer-space
// shape a decision model like Microsoft-Decision-1 expects.
const res = await fetch("https://jevtypesafeai.com/api/v1/decide", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    Authorization: `Bearer ${process.env.JEV_KEY}`, // jv_live_...
  },
  body: JSON.stringify({
    state: { action: agent.proposedAction },
    questions: {
      safe: { type: "noul", instructions: "Is this action safe to run without review?" },
      area: { type: "choice", instructions: "Which system does it touch?",
              criteria: { billing: "money", data: "user data", infra: "servers" } },
    },
  }),
});
const { answers } = await res.json();
if (answers.safe.noul > 0.85) run(); else review(answers.area.choice);

• Digital Today — Microsoft joins the decision-focused AI model race with Microsoft-Decision-1 — Task types, Qwen3.5-9B backbone, Foundry/OpenRouter availability, and per-team speed/cost claims (vendor-reported).
• Crypto Briefing — Microsoft launches Decision-1 — Nadella's announcement framing and positioning in the decision-model category.

FAQ

See also: Microsoft-Decision-1 vs Jev · What is a decision model · Clef — Cloudflare's decision models · Jev alternatives · Playground