Kimi-K3 takes the top spot on the frontend code leaderboard; China is awesome! Feels like frontend code has become the breakthrough point for domestic models. But I think the ultimate competition in AI is still about the battle for entry points—whoever can secure the entry point, integrate into workflows, shape user habits, and build a closed-loop ecosystem will truly come out on top. Here’s my current experience: Code is awesome: Kimi/Claude/GPT Writing logic is awesome: Claude/GPT Math is awesome: OpenAI Chinese scenarios are awesome: Kimi/DeepSeek Search is awesome: Perplexity Each has its strengths.