GitHub Copilot Adds Kimi K2.7 Code, Its First Open-Weight Model

GitHub Copilot just added Kimi K2.7 Code to its model picker, and the detail that matters isn't the model itself, it's the precedent. This is the first open-weight model Copilot has ever offered as a selectable option, sitting alongside proprietary models from OpenAI, Anthropic, and Google. For a tool used by millions of developers, that's a quiet but real shift in how mainstream coding assistants treat model choice.

What Kimi K2.7 Code actually is

Kimi K2.7 Code comes from Moonshot AI, a Beijing-based lab, and it's open-weight rather than fully open-source, meaning the model weights are published even if the training data and code aren't. According to GitHub's changelog, it went generally available in the Copilot model picker on July 1, 2026, hosted on Microsoft Azure rather than Moonshot's own infrastructure. Technically, it's a mixture-of-experts model with roughly 1 trillion total parameters, but it only activates about 32 billion of them per token, which is what makes it cheap to run relative to its apparent scale. GitHub bills usage at provider list pricing rather than a flat Copilot subscription rate.

Why GitHub is rolling it out carefully

The rollout has been staged. Kimi K2.7 Code first reached Copilot Pro, Pro+, and Max subscribers, then GitHub extended it to Copilot Business and Enterprise plans on July 7, 2026. For those organizational tiers, it ships off by default: an admin has to flip on the Kimi K2.7 Code policy in Copilot settings before anyone on the team can select it. That's a sensible guardrail, since enterprise customers tend to have stricter requirements around where code and prompts get processed, even when a model's weights are open.

The bigger pattern: open-weight models catching up on price and capability

This launch doesn't happen in a vacuum. Chinese open-weight models are increasingly cited as running 60 to 90 percent cheaper than frontier models from Anthropic and OpenAI, and models like Z.ai's GLM-5.2 have reportedly seen some of the fastest adoption curves of any model released in 2026. Enterprise API traffic flowing through US developer platforms is now estimated to include a meaningful share running on Chinese-built models, largely because the price-to-capability ratio has become hard to ignore for everyday coding tasks. Kimi K2.7 Code landing inside Copilot, arguably the most mainstream AI coding tool that exists, is a sign that this price pressure is no longer confined to niche developer forums or specialist API platforms. It's showing up in the default tool millions of engineers already have installed.

The catch is that "open-weight" and "cheap" don't automatically mean "as capable" for every task. Frontier proprietary models still tend to lead on the hardest reasoning and long-context work, and enterprises weighing a switch have to account for hosting, data handling, and support commitments, not just per-token price. What Copilot's move really signals is optionality: developers can now compare an open-weight model against GPT- and Claude-class options on the same task, in the same interface, and decide per query whether the frontier model's edge is worth the extra cost. That same logic, letting the model compete on a case-by-case basis rather than committing to one vendor, is the premise behind bring-your-own-key tools like ByteChat, which let you run multiple providers side by side instead of picking one.

Frequently asked questions

What is Kimi K2.7 Code?

It's an open-weight AI coding model from Moonshot AI, a Beijing-based lab, built as a mixture-of-experts model with about 1 trillion total parameters but only 32 billion activated per token, which keeps its per-query compute cost relatively low.

Is Kimi K2.7 Code available to all GitHub Copilot users?

It's live for Copilot Pro, Pro+, and Max plans, and as of July 7, 2026, it's available to Copilot Business and Enterprise too, though it's off by default and an admin must enable it for the organization.

Why does it matter that GitHub added an open-weight model?

It's the first open-weight model in Copilot's model picker, showing that the price gap between cheap open-weight models and expensive frontier models has gotten large enough that even mainstream tools are giving developers the choice.

Whether that choice tilts toward the frontier model or the cheaper open-weight one will likely keep changing every few months.