Microsoft is testing Microsoft-Decision-1, a model built for scoring decisions, in both Copilot and Bing Search, Search Engine Land reported on October 12. Microsoft pitches it for routing and classification, prioritization, verification, and steering workflows inside apps and agents. The company claims it outpaces LLMs and rival decision models on latency and quality, but the coverage cites no benchmark.
Rangan Majumder, Microsoft’s VP of Copilot Quality, said on X that the model is “pretty good” in Copilot. Krishna Madhavan, a Principal Product Manager for Bing at Microsoft AI, added on X that “we are always experimenting!” That confirms a test, not a launch, and nobody named the queries or users involved.
Neither the announcement nor the two staff comments explain how Bing applies the model, so any ranking claim is unsupported. The more useful reading: Microsoft is placing small scoring layers beside its large language models, the kind of component that can decide which step or source an answer system takes next. SEO teams should log Bing and Copilot behavior now, so any later shift has a baseline.
Search Engine Land, reported by Barry Schwartz on October 12, 2026.
