Moonshot AI's Kimi K3 model has achieved a 'DeepSeek moment' and is now being seriously evaluated by Microsoft for use...

Moonshot AI's Kimi K3 model has achieved a 'DeepSeek moment' and is now being seriously evaluated by Microsoft for use in Copilot, according to multiple reports. The Financial Times profiles founder Yang Zhilin, noting that Kimi K3 has closed the gap with leading US rivals. Separately, The Information reports that Microsoft is internally testing Kimi K3 to handle some Copilot inference requests currently processed by OpenAI and Anthropic models, estimating potential annual cloud infrastructure cost savings of up to $600 million. This reflects Microsoft's strategy to diversify its AI model suppliers and control costs. While no final decision has been made, the evaluation covers coding, reasoning, and cost efficiency. Any deployment would face technical integration, data-sovereignty, and export-control hurdles, and may initially be limited to less sensitive workloads. This development signals a potential design win for Moonshot AI, establishing it as a serious competitor in the enterprise AI market.

What the sources said:

• FT: 'Yang Zhilin, founder of Moonshot AI, faced early doubts over revenue and model capabilities before its Kimi K3 delivered a “DeepSeek moment.”' (Financial Times)
• The Information (via TechNode): 'Microsoft estimates that shifting part of Copilot’s workload to Kimi K3 could reduce annual cloud infrastructure costs by up to $600 million.' (TechNode)
• The Information (via TechNode): '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.' (TechNode)