Microsoft enters decision-model race with Qwen-based offering
Microsoft introduced Microsoft-Decision-1, initially built on Qwen3.5-9B from a Chinese AI lab, to compete in the growing decision-model category. The company says a later version will use its own technology. The move follows TypeSafe AI's Jev and a wave of similar decision models from OpenAI, Cloudflare, Liquid AI, Perplexity, and others.
Microsoft's new decision model, Microsoft-Decision-1, is initially powered by Alibaba Cloud's Qwen3.5-9B. Redmond says a future release will shift to its own models and OpenAI technology. It is available through Microsoft Foundry and will later appear on OpenRouter.
The decision-model field has grown crowded. TypeSafe AI's Jev helped spark interest, while OpenAI, Cloudflare, Strands, Liquid AI, Perplexity, Snowflake, Surogate, and H2O.ai have launched or promoted similar tools. Microsoft claims advantages in speed, benchmark accuracy, confidence, and text-classification cost, with free output tokens and paid input tokens.
Businesses and developers may gain cheaper, faster ways to automate classification and routine choices, potentially reducing reliance on general-purpose chatbots. Workers whose tasks involve triage, support, or document sorting could see workflows change, though human oversight may remain important because decision models can still err. As more vendors enter, buyers could face both more choice and harder comparisons around accuracy, latency, and cost.