Moonshot AI launches Kimi K3, a 2.8T parameter open-weight model

Kimi K3 has 2.8 trillion parameters, a 1 million token context, and native multimodal capabilities. It introduces Kimi Delta Attention for up to 6.3x faster decoding and Attention Residuals for ~25% higher training efficiency. The model is now available on Nebius and via open weights.