Microsoft-Decision-1

Microsoft-Decision-1, announced October 9, 2026 by Achint Srivastava (VP of Software Engineering, Office of the CTO) and unveiled by Satya Nadella, is Microsoft's entry in the decision-model wave: a purpose-built model that takes a state description and a fixed set of answer options and returns a calibrated probability for each, instead of generating text.

Microsoft post-trained Alibaba's open-weight Qwen3.5-9B for single-pass decision scoring, and plans to rebase the system on other foundations including Microsoft AI (MAI) and OpenAI models. It supports yes/no questions, multiple-choice selections, ratings, and rubric-based grading of AI responses or agent actions, over a 32,768-token text context with JSON output. Pricing is $0.042 per million input tokens with output free. The weights are not released — hosted API only.

Microsoft's claims are aggressive: top accuracy across 36 benchmarks (~150,000 held-out questions), 4.5x faster than runner-up Quyet-1.0-Large, and 35x faster than GPT-6 Sol at a P50 of 85 ms, with decisions flipping only 1.3% of the time under perturbations. The independent check is more modest: a third-party bench on October 10 scored it 75/82 correct (6th of 14 decision models) at 0.472 s median over the network — about 2.4x faster than GPT-6 Sol, not 35x — with 0/26 option-order flips but a miss on a prompt-injection test other cloud models resisted. It is available in Microsoft Foundry and via OpenRouter as microsoft/microsoft-decision-1.

Provider-reported: top of 36 benchmarks, P50 85 ms, 35x faster than GPT-6 Sol. Independent third-party bench (Oct 10, 2026): 75/82 correct (6th of 14), 0.472 s median — ~2.4x faster than GPT-6 Sol; 0/26 option-order flips.