Two days after release, Moonshot AI's Kimi K2 Thinking reached the top of the Hugging Face open-source rankings, scoring 44.9% on the HLE benchmark and surpassing Grok 4, GPT-5 and Claude 4.5, with improved agentic search, coding and comprehensive reasoning. Its API pricing is well below closed-source flagship models.
Informed sources put the training cost at around 4.6 million US dollars - lower than DeepSeek V3's 5.6 million. A Hugging Face co-founder described it as 'another DeepSeek moment,' highlighting the continued momentum of China's low-cost open-source model route.
