Kimi K3 is now available in GitHub Copilot
Kimi K3 is now generally available in GitHub Copilot. The open-weight model is hosted by GitHub and was announced through the GitHub Changelog and The GitHub…
By Dillip Chowdary • Aug 06, 2026 • Source: GitHub Changelog
Kimi K3 is now generally available in GitHub Copilot. The open-weight model is hosted by GitHub and was announced through the GitHub Changelog and The GitHub Blog.
On the technical side, Kimi K3 is positioned for agentic coding. Availability inside Copilot means developers can call it through the same Copilot surface they already use, without running or hosting the open-weight weights themselves. GitHub hosts the model, so inference and routing sit behind Copilot rather than in a separate vendor console.
Advertisement
Tech Pulse Daily
Get tomorrow's pulse first
Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.
For engineers and builders, the useful part is model choice inside a familiar coding assistant. Teams that care about agentic coding workflows can try a frontier-class open-weight option without changing tools or wiring a new API path. Hosting by GitHub also keeps model selection inside one product surface instead of spreading it across multiple providers.
In market terms, this puts an open-weight model with frontier-level agentic coding ability and highly cost-effective pricing directly into a mainstream IDE assistant. That tightens the gap between closed frontier models and open-weight options for day-to-day coding work, and it gives Copilot another competitor-grade model under its own hosting.
Practical takeaway: if you already use GitHub Copilot, treat Kimi K3 as a selectable model for agentic coding tasks and compare quality versus cost against your current default. Watch how often you prefer it for multi-step coding work, and whether the cost profile holds up under real team usage.
Advertisement
🔎 More interesting news
- Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users
- Qwen 3.8-Max and Claude Opus 5 show why raw benchmark scores don't predict the bill
- Loop Engineering with native model switching in Codex and Claude
- Show HN: Reduck GEO, open source Skill to measure and optimize Claude citations
- Today's full Tech Pulse briefing →