Microsoft launches Decision-1, an AI model built to make fast calls

Microsoft has unveiled Microsoft-Decision-1, an AI model with a narrow job: making structured decisions quickly. It doesn’t write poems or draft emails. It picks the right answer from a fixed list.

Microsoft says the model tops rivals on accuracy and runs far faster than them. The company is already testing it inside its own operations.

What Decision-1 actually does

The model was announced on October 9, 2026. Microsoft describes it as a decision-scoring model built for tasks like routing, classification, prioritization, verification, and workflow control.

Technically, the model assigns calibrated probability scores to a set of fixed answer options. Those can be simple yes/no questions or multiple-choice setups. Calibration matters because it means the confidence number is supposed to be trustworthy, not just decorative.

Plainly put, the model doesn’t just say “route this ticket to security.” It also says how sure it is. That lets a business decide when to trust the machine and when to escalate to a human.

Decision-1 can handle inputs of up to 32,000 tokens in a single pass. A single-pass design means the model takes in the whole input and returns its scores in one shot, rather than generating an answer word by word.

The numbers Microsoft is pointing to

Microsoft claims Decision-1 achieved the highest accuracy across a benchmark suite of nearly 150,000 blind questions.

According to Microsoft, the model’s latency is 4.5 times faster than Quyet-1.0-Large, which the company identifies as the nearest competitor.

Against GPT-6 Sol, the gap reportedly widens to 35 times faster.

These are Microsoft’s own figures. Independent testing will be the real proving ground, as it is for any vendor benchmark.

Built on someone else’s foundation

Decision-1 is not built on a Microsoft-made base model. It is a post-trained version of Qwen3.5-9B, the open-weight model from Alibaba.

Microsoft has said future updates are planned on Microsoft AI (MAI) and OpenAI models.

The 9B in the name refers to the model’s size. It sits in the compact category, which helps explain the speed claims. Smaller models generally need less computing power per answer.

Where it’s being tested and where it’s going

Microsoft says Decision-1 is currently in internal trials across incident response, quality control, and scientific discovery workflows.

For outside users, the model is available through Microsoft Foundry and is scheduled for deployment on OpenRouter. Foundry is Microsoft’s platform for building and running AI in enterprise settings. OpenRouter gives developers a single gateway to many different models.

The broader shift toward smaller, specialized AI

Decision-1 does not arrive alone. Similar specialized offerings from Liquid AI and vLLM Semantic Router have emerged in the same time frame.

These models are designed to handle high-volume tasks more efficiently than general-purpose models that were never built for rapid-fire classification.

What this means

The calibrated scoring may be the most practical feature. Companies deploying AI in sensitive areas need to know when a model is unsure. A confidence score creates a clean handoff point between automation and human review.

The benchmark results are self-reported, and the testing remains internal for now. The things to watch: the OpenRouter rollout, independent benchmark comparisons, and whether Microsoft reports results from its incident response and quality control trials.