Kimi K3: China's Open-Source World-Class Model – What Makes It Strong?

Moonshot AI has released a new open-source large language model, Kimi K3, which the developers claim can compete with world-class models such as GPT-4o, Claude 3.5 Sonnet, and DeepSeek V3. The model boasts high performance, achieving outstanding results on several benchmarks, including AIME, MATH-500, and GPQA. Kimi K3 supports processing long contexts up to 128K tokens, multiple image recognition in a single response, and high inference speed, generating up to 120 tokens per second. The model is available for commercial use with open source, allowing developers to integrate it into their projects.