Chinese AI Surges: Kimi K3 Rivals OpenAI as Google Delays

Google's Gemini 3.5 Pro has been delayed due to performance issues. Meanwhile, Chinese AI companies are unveiling new models that are rapidly catching up to leading US systems. Moonshot AI recently introduced Kimi K3, a 2.8-trillion-parameter model. Alibaba also previewed its 2.4-trillion-parameter Qwen 3.8 Max. Kimi K3 has shown benchmark results that are close to Anthropic's Fable 5 and OpenAI's GPT-5.6. On some benchmarks, it even outperformed OpenAI's highest-end reasoning model, GPT-5.6 Sol. Analysts note that the idea of Chinese AI models lagging US models by six months is becoming harder to support. Kimi K3 scored third in the Artificial Analysis Intelligence Index, an independent benchmark of AI models. It scored 57, placing it behind only Fable 5 and GPT-5.6. It surpassed Gemini 3.0 Pro, Claude Opus 4.8, and others. The model performed very well in agentic AI and coding. It ranked fourth in AA's Coding Agent Index. Kimi K3 also ranked first on Frontend Code Arena, beating GPT-5.6. It showed strong results on several other agent benchmarks, including surpassing GPT-5.6 Sol on τ³-Banking. This shift means the global AI landscape is becoming more competitive and diverse.