Microsoft (Nasdaq: MSFT) is testing Moonshot AI‘s Kimi K3 on Azure to see whether the Chinese open-weight model can handle Copilot tasks OpenAI and Anthropic systems currently run, according to reporting from The Information.
The switch could save Microsoft up to $600 million in inference costs. Microsoft has not confirmed the figure or named which Copilot tasks might move. Kimi K3 is currently still absent from Microsoft’s Foundry pricing page, which lists only older Moonshot models, suggesting the Azure integration remains unfinished.
there it is – microsoft will use kimi k3 to power their flagship copilot ai assistant saving up to 60% per token or $600M on every $1B spent on inference spend.
— Ejaaz (@cryptopunk7213) July 20, 2026
if microsoft chooses to self-host the model once Kimi open sources it on the 27th then this cost reduces even more… pic.twitter.com/BwrXXZvAim
Beijing-based Moonshot AI released Kimi K3 on July 16 as a 2.8 trillion-parameter multimodal model with a one-million-token context window, built for coding, knowledge work and reasoning. It topped LMArena’s coding leaderboard, undercut OpenAI on price, and drew enough demand that Moonshot paused new K3 subscriptions over compute strain. Moonshot plans to release the full weights on July 27, letting developers self-host and customize the model instead of relying on the API alone.
Microsoft engineers must confirm K3 meets Copilot’s requirements for quality, reliability, safety and latency before any feature moves off OpenAI or Anthropic. Microsoft ran similar internal checks on DeepSeek and earlier Kimi releases without switching, a person with knowledge of the plans told The Information, indicating the company treats this kind of test as routine.
GitHub Copilot has offered Kimi K2.7 Code since July 1, Azure-hosted with enterprise admin controls that default off. Microsoft has pushed a multi-model strategy through Foundry, positioning the platform as model-agnostic and pointing developers toward capability, safety, latency, and cost over defaulting to the most powerful system for every task.
The Trump administration is reportedly weighing an executive order or other measures restricting Chinese open-weight AI models in the US. Microsoft, OpenAI’s largest financial backer and an Anthropic partner, is testing a Chinese competitor’s model for a flagship product just as Washington considers cutting off that option.
JUST IN: 🇺🇸🇨🇳 Trump administration weighs strict curbs on Chinese AI models after Moonshot Kimi K3 launch, including U.S. host liability and backlist threats. pic.twitter.com/tJrnwKBgpm
— Whale Insider (@WhaleInsider) July 20, 2026