Kimi K3 Fast

by Moonshot

Kimi K3 Fast is Moonshot's Kimi K3 delivered through a speed-optimized serving tier rather than a different model. The weights are unchanged, a Mixture-of-Experts design with 2.8 trillion total parameters and 104 billion active per token, released July 2026 under the Kimi K3 License and built for long-horizon coding, knowledge work, and reasoning. The tier keeps the full 1,048,576 token context window and output cap, which is what makes it useful for the work K3 is built for, whole-repository analysis and extended agent runs where the reasoning trace itself grows large. K3 runs in preserved thinking history mode, so multi-turn calls should pass complete assistant messages back including their reasoning content rather than stripping it. The trade is straightforward, faster completions at a higher per-token price than the standard K3 route, so it fits interactive coding agents and tool loops where latency is the binding constraint, while batch and offline jobs are better served by the standard tier.

Key info

Input
Output
Features
Context window
1M
Max output
1M

Available routes

No routes currently available — Kimi K3 Fast isn't routed through the Opper gateway right now. It may return.

Contact us about this model →

Available models from Moonshot

Start building with 700+ models

One API key. Every major provider. Up and running in minutes.

Get startedView Documentation