GitHub Copilot Opens to All Plans and Gets Its First Open-Weight Model — Kimi K2.7 Code Now in Enterprise

TL;DR: Kimi K2.7 Code — the first open-weight model in GitHub Copilot's model picker — expanded to Copilot Business and Enterprise this week, alongside a new AI credit pool feature that caps agentic spending per cost center.

Moonshot AI's Kimi K2.7 Code (1T total parameters, 32B active per token via a mixture-of-experts architecture) is now selectable in GitHub Copilot for Business and Enterprise plans. It's hosted by GitHub on Microsoft Azure and billed at provider list rates under usage-based billing — admins need to enable the policy before users can pick it.

In the same window, GitHub launched AI credit pools: admins can set hard spending caps scoped to cost centers, addressing a real concern as agentic coding tasks consume roughly 1,000× more tokens than single-turn chat. The free Copilot desktop app also expanded to all plan tiers including Copilot Free, with each session running in an isolated git worktree.

Why it matters: Enterprise teams can now run an auditable open-weight model through a managed SaaS tool — getting the model-weight transparency of open-weight AI without operating their own inference infrastructure.

Sources: Kimi K2.7 in Copilot Business/Enterprise — GitHub Changelog · Copilot desktop app for all plans — GitHub Blog