Kimi K2.7 Code Becomes First Beijing Model Available Inside GitHub Copilot

Moonshot AI's Kimi K2.7 Code is now generally available inside GitHub Copilot — making it the first open-weight model from a Beijing-based lab to be selectable in the world's most widely used AI coding assistant.

The speed is the story. Moonshot published K2.7 Code's weights on Hugging Face, and GitHub made it available inside Copilot within 19 days — one of the fastest transitions from open-weight release to enterprise-platform availability on record.

Why it matters: The integration is a milestone for Chinese open models' penetration of Western developer tooling. Open weights and aggressive pricing let Kimi undercut closed rivals, and Copilot's vast user base gives it instant distribution that Chinese labs have historically struggled to reach.

The caveat: As of early July, no independent third-party benchmark results for Kimi K2.7 Code existed on any public leaderboard — all performance figures come from Moonshot's own proprietary benchmark suites. And the model "audits differently" than Western alternatives, a nod to the data-governance questions that still shadow Chinese-origin models in regulated enterprises.

Kimi's Copilot debut sits alongside a broader wave — GLM-5.2, DeepSeek V4, MiMo — of Chinese open models winning real adoption in the tools developers actually use. Moonshot, meanwhile, is reportedly readying a far larger Kimi K3 with a rumored 2.5 trillion parameters for later this year.