Unverified50% confidenceBenchmarkExact time
Kimi K3 on AMD MI355X using grouped-head MLA verification kernels achieves 1.37–1.77× throughput improvement and 1.45–2.42× ITL improvement at concurrencies of 2–32.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/5/2026
First Seen
Valid until: 12/4/2026
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.66% similarUnverifiedKimi K3使用自研的Kimi Delta Attention机制，解码速度最高提升6.3倍，上下文窗口支持100万token63% similarUnverifiedKimi K3 was released on July 16 and integrated into Kimi Code, featuring a parameter scale of 2.8 trillion, supporting up to 1 million token context and native visual understanding.62% similarUnverified混元3是混合专家模型（MoE），总参数量为295B，激活参数约21B61% similarUnverifiedKimi's parameter count grew from 1T in K2 to 2.8T in K3.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/860554API
curl https://kongchang.com/api/v1/knowledge/claims/860554MCP
get_claim(id=860554)