moonshotai / Kimi-K3

Vision2780B813.1K11.9KUpdated Sep 2, 2026

Downloads

813.1K

Likes

11.9K

Last updated

Sep 2, 2026

License

kimi-k3

Kimi K3 is an open-weight, native multimodal agentic model and our most capable model to date. It is a 2.8T-parameter model built on Kimi Delta Attention (KDA) and Attention Residuals (AttnRes), with native vision capabilities and a 1-million-token context window. It is the wo…

Tasks & library

  • image text to text
  • transformers

Tags

  • transformers
  • kimi_k3
  • feature-extraction
  • compressed-tensors
  • conversational
  • image-text-to-text

Technical specifications

Parameters2780B
Librarytransformers
LanguagesNot available
Licensekimi-k3