Microsoft is internally testing Moonshot AI’s Kimi K3 and evaluating whether some Copilot inference requests currently handled by OpenAI and Anthropic models could be moved to Kimi K3, The Information reported. Microsoft has not made a final decision on replacing any models.
Microsoft estimates that shifting part of Copilot’s workload to Kimi K3 could reduce annual cloud infrastructure costs by up to $600 million. The evaluation reflects Microsoft’s broader effort to use multiple AI models and reduce its dependence on a single supplier while controlling the cost of serving large volumes of AI requests.
Kimi K3 is reportedly being assessed not only for its performance in coding and reasoning, but also for its cost efficiency in specific Copilot tasks. Any deployment would still involve technical integration, data-sovereignty, security, and export-control considerations, and could initially be limited to less sensitive workloads. [The Information]