Microsoft launches Microsoft-Decision-1 judgment AI at $0.042 per 1M tokens

Microsoft-Decision-1 arrives in a niche that drew attention in September. TypeSafe AI released Jev on September 15, and OpenAI previewed its Decisions API for the low-cost model Luna on September 29 — both aimed at the same kind of option-scoring role that Decision-1 fills.

Microsoft is entering at the same price as Jev: $0.042 per 1M input tokens with free output. The model is built on Alibaba's Qwen3.5-9B with additional Microsoft training, and Microsoft says it plans to move to bases built on Microsoft AI or OpenAI models later. It is available through Microsoft Foundry and OpenRouter.

Microsoft's own testing put Decision-1 first on accuracy across 36 benchmarks it says were not used in training, and fastest in response speed — 2.5 times quicker than runner-up H2O-Lightning-4B v1.1 and 35 times quicker than GPT-6 Sol. On calibration, Microsoft's comparison chart placed it third behind Jev and others; it later updated its blog to add accuracy and calibration comparisons with Jev. Microsoft also cites internal use by an Xbox research team, which classified over 10,000 pieces of game-related feedback at quality comparable to GPT-6 Sol while running more than 14 times faster at 1/200th the cost.