Intel / article

📣 @Kimi_Moonshot's Kimi K3, an open-weight model, is now generally available an

Source: github.blog
Author: GitHubTop Viber
Date: Published Aug 6, 2026

Context

GitHub added Kimi K3 to Copilot as a model GitHub runs, so its do not make this a local or self-hosted option. Kimi K3 comes from Moonshot AI and is an open-weight model, meaning its trained weights are published for others to download and run. In Copilot, GitHub's August 6 changelog says the model is hosted by GitHub on Fireworks AI, so requests are processed by that hosted service rather than on the user's machine. GitHub calls its agentic coding abilities frontier-level and its pricing highly cost-effective, but the post gives no or comparison behind that claim.

Access is opt-in for organizations and billed per . GitHub said the rollout would be gradual across Copilot Pro, Pro+, Max, Business and Enterprise. For Business and Enterprise, the model is off by default and stays unavailable until an administrator enables the Kimi K3 policy in Copilot settings, and GitHub recommends reviewing open-weight models against security, compliance and data-governance requirements before doing so. The model is billed at provider list pricing under usage-based billing, which an editor's note gave as $3 per million input tokens, $15 per million output tokens and $0.30 per million cached input tokens. Two editor's notes, both dated August 6, record that GitHub paused the rollout during a GitHub Actions incident and then resumed it. GitHub's current models and pricing documentation lists the same three rates for Kimi K3.

Terms in this piece · Glossary

open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.

Read the source github.blog