Microsoft's Decision-1 model enters the fast-growing AI decision model race

You can now use a more accurate and faster AI model for common routing and classification tasks without changing your existing workflows.

Microsoft has launched Decision-1, a new AI model designed for fast, structured decision-making tasks such as classification, evaluation, and routing. Built on the Qwen3.5-9B architecture, Microsoft claims Decision-1 outperforms competitors across 36 benchmarks, achieving 83.5% accuracy and 85 ms latency. It is positioned to guide and control AI agents in complex environments. Decision-1 is available via Microsoft Foundry and OpenRouter, with pricing at $0.042 per million input tokens and free output tokens. This release enters a rapidly expanding market for decision models, a trend initially sparked by startups like Jev.

Microsoft's entry with Decision-1 signifies a maturing market for specialized AI decision models, moving beyond early entrants like Jev. The availability on platforms like Microsoft Foundry and OpenRouter suggests a push for broader adoption and integration into existing AI agent frameworks. This competition is likely to drive down costs and increase performance across the board for decision-making AI.

The model's performance claims, particularly its speed and accuracy on benchmarks, indicate a significant leap in practical AI application. For developers building autonomous agents or complex workflow automations, Decision-1 offers a potentially more robust and efficient tool, setting a new bar for performance in this AI sub-field.