Search
The Lunar Shadow's Edge: Kimi K2 Thinking's Cognitive Prowess Outshines GPT-5
4 day ago /
Read about 0 minute
Author：小编
The latest innovation from The Lunar Shadow, the Kimi K2 Thinking model, has etched its name in the annals of artificial intelligence by securing a remarkable 44.9% score in the Human Cognitive Benchmark Exam (HCB), outperforming cutting-edge models like Grok4, GPT-5, and Claude 4.5. The Kimi K2 Thinking Pro variant takes it a step further, boasting an impressive 51% score. A co-founder of HuggingFace remarked that this milestone reaffirms the ascendancy of open-source solutions in the AI landscape. Insider reports reveal that the model's training incurred a cost of $4.6 million, a figure that pales in comparison to the expenditures associated with DeepSeek V3 and OpenAI's GPT-3.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
