Unverified50% confidenceFactExact time
Kimi K3 is a 2.8T-parameter multimodal LatentMoE model that uses 896 experts with top-16 routing and performs routing computation in a 3584-dimensional latent space.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedKimi K3's 2.8 trillion parameter scale surpasses the previously largest open-weight model, DeepSeek-V3, which had 671 billion parameters.76% similarVerifiedKimi K2.6 adopts a Mixture of Experts (MoE) architecture, with a total parameter count at the trillion level and only 32B parameters actually activated.71% similarUnverified以Kimi K3实验，给它预填充1%的Claude Opus解码推理片段，其输出风格就会明显向Opus靠拢，某些区段Claude推理链从Kimi K3提取的容易程度比其他模型高出约6个数量级71% similarUnverifiedKimi's parameter count grew from 1T in K2 to 2.8T in K3.70% similarUnverifiedKimi K2.6 supports task chain execution of more than 4,000 steps.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/897548API
curl https://kongchang.com/api/v1/knowledge/claims/897548MCP
get_claim(id=897548)