Microsoft is evaluating Moonshot AI's Kimi K3 model for potential use in Azure Copilot, the company's AI-powered coding assistant. The move comes as the tech giant looks to diversify its AI suppliers and manage costs.
Why the price matters
Kimi K3 recently topped a coding benchmark with 1,679 points, outperforming several competing models. More notably, it underbid OpenAI on price, offering a cheaper alternative for the same task. That cost advantage could be a key factor for Microsoft, which currently relies heavily on OpenAI's models for Copilot and other services.
Microsoft has invested billions in OpenAI and uses its GPT models across products like Azure, Office, and Bing. But the company has also been exploring other AI options. Testing Kimi K3 suggests Microsoft is serious about having backup suppliers, especially as the AI market heats up.
What the evaluation means for Azure Copilot
Azure Copilot helps developers write and debug code. If Microsoft decides to integrate Kimi K3, it could lower operational costs and reduce dependency on a single provider. The benchmark score indicates the model is competitive on technical tasks, though real-world performance may vary.
Moonshot AI, a relatively young company, has been gaining attention for its efficient models. The evaluation is still in early stages, and no timeline for a decision has been disclosed.
For now, Microsoft continues to use OpenAI's models as the primary engine for Copilot. But the company's willingness to test alternatives signals a shift in strategy — one that could reshape the AI assistant market.
