kimi-k3

kimi-k3 is Kimi's flagship model, built for long-horizon coding and end-to-end knowledge work, with native visual understanding. K3 always runs in thinking mode. You can set the thinking effort via the top-level reasoning_effort parameter (supports low, high, max; defaults to max). It supports a context window of up to 1M tokens.

LiveMoonshot2 protocols1M contextStream cancellation unsupported