Microsoft launches Decision-1 on October 9, post-trained on Qwen3.5-9B, claiming 35x faster decisions than GPT-6 Sol

Microsoft launched Microsoft-Decision-1 on October 9, the company's first model purpose-built for fast, structured decision-scoring tasks, entering the same lane as OpenAI's Decisions API and TypeSafe's Jev. It is post-trained on Alibaba's open-weight Qwen3.5-9B and targets high-frequency, low-latency tasks such as routing, classification, and intent recognition rather than open-ended conversation. Microsoft claims the model scored highest on a 36-benchmark comparison of nearly 150,000 questions and is 35 times faster than GPT-6 Sol on decision tasks, with input pricing of $0.042 per million tokens. Microsoft-Decision-1 is now available in Microsoft Foundry and will soon be available through OpenRouter. The launch continues Microsoft's pattern of post-training on Qwen open weights and creates direct competition with OpenAI's October 6 Decisions API public beta on GPT-6 Luna and TypeSafe AI's September 15 launch of Jev. Microsoft said the model enters the decision-task market as a specialized offering, giving Azure AI Foundry customers a new routing and classification option; the 35x speed figure is Microsoft's own claim, and no independent third-party benchmarks have been published.