Moonshot AI's Kimi K3 tops global AI leaderboard with 2.8 trillion parameters

Moonshot AI recently launched the Kimi K3 model, boasting 2.8 trillion parameters and a 1 million token context window. It topped the global Frontend Code Arena leaderboard with a score of 1679, surpassing Claude Fable 5.

This achievement has caught industry attention. UC Berkeley professor Ion Stoica noted that the gap between Chinese open-source models and the most advanced ones has narrowed from 6-9 months to just 2-3 months. Kimi K3's performance has also sparked debate about the US AI lead, with some businesses now using 'model routing' services to pick the most cost-effective AI for tasks, including Chinese models.

US AI executives say they must clearly prove their products' value to customers as competition heats up. OpenAI CEO Sam Altman previously acknowledged that Chinese open-source models have become very strong.