Unverified50% confidenceFactExact time
The research team used Kimi K3 as the LLM judge, whose scores were highly consistent with human annotations.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/11/2026
First Seen
Valid until: 12/10/2026
Sources
Related Entities
Related Claims
UnverifiedThe evaluation uses three scoring dimensions: Instruction Following, Unit Testing, and LLM as Judge62% 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token62% 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.62% 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.61% similarUnverifiedClaude 3.7 Thinking was used as the LLM judge in the evaluation, providing high consistency with minimal variance across repeated runs60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/893449API
curl https://kongchang.com/api/v1/knowledge/claims/893449MCP
get_claim(id=893449)