2.8 trillion parameters. 1 million token context. Native multimodal.


Kimi Delta Attention delivers up to 6.3x faster decoding at million-token context, cutting KV-cache memory by up to 75% along the way.
Attention Residuals deliver roughly 25% higher training efficiency at under 2% additional compute cost.
Built for long-horizon agentic coding and self-evolving workflows, sustaining long engineering sessions with minimal human oversight.
post-image
post-image
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned