Microsoft is strengthening its AI product lineup with a specialization strategy.

On Friday, Microsoft CEO Satya Nadella announced the launch of a new-generation fast decision AI model, Microsoft-Decision-1, which is specifically designed for structured decision-making tasks.

Nadella posted on the X platform that the model "performs excellently on structured decision tasks, outperforming large language models and other decision models in both latency and quality," and has already been tested internally at Microsoft for scenarios including incident response, quality control, and scientific research.

Microsoft stated that the model runs 35 times faster than OpenAI's GPT-6 Sol and 4.5 times faster than Quyet-1.0-Large in internal tests.

In terms of pricing, the model charges $0.042 per million tokens for input, while output is free, a structure that is clearly attractive at the cost level for high-frequency, repetitive decision applications.

Focused on decision-making rather than generation, filling the capability gap of LLMs

Unlike traditional large language models that emphasize text generation and complex reasoning, Microsoft-Decision-1 has a more focused design goal.

The Microsoft-Decision-1 model selects the optimal solution from preset options and assigns probability scores to each candidate answer, thereby helping downstream applications determine whether to continue execution, retry, escalate processing, or hand it over to human review.

Microsoft stated that the model achieved the highest accuracy across 36 benchmark tests covering nearly 150,000 questions. Its functional positioning includes routing and distribution, content classification, task prioritization, result verification, and workflow control, and it can be seamlessly embedded into existing applications, AI agents, and workflow systems.

However, the above performance data comes from Microsoft's own benchmark tests, and independent third-party verification has not yet been completed.

Internal tests show significant speed and cost advantages

Multiple internal teams at Microsoft have tested the model in real-world scenarios.

The Xbox Research team used it to classify more than 10,000 pieces of game feedback, and the results showed comparable quality to GPT-6 Sol, but with more than 14 times the speed and about 200 times lower cost. The Microsoft Copilot team also found that the model performed similarly to GPT-5.6 Luna on AI response evaluation tasks.

In terms of technical architecture, Microsoft-Decision-1 is built on Qwen3.5-9B and has undergone specialized post-training centered on single-decision scoring.

Microsoft also plans to migrate the model in the future onto other foundations such as its MAI series models and OpenAI models.

In the announcement, Microsoft emphasized that the model is intended to "integrate decision intelligence into existing applications in a safe and trusted environment," positioning it as one of the foundational components for AI agent workflow orchestration.

Currently, Microsoft-Decision-1 has been opened to developers through Microsoft's AI development platform Microsoft Foundry, and support for access via OpenRouter is also planned.