China just passed Claude Opus with Kimi K3

Kimi K3 just passed Claude Opus 4.8 on Artificial Analysis's benchmark index and took #1 on the Arena WebDev coding leaderboard — at the exact same API price as Claude Sonnet 5. Moonshot AI's new 2.8-trillion-parameter open-weights model is the first open model priced like a frontier one. This is a plain-numbers read of the Kimi K3 launch for people who pay for AI at work: what the benchmarks actually show (including the knowledge-work results most coverage skipped), why the price is the real story, the honest catch (hallucination rate went up, and reasoning costs add up), and what this launch does to your Claude or ChatGPT bill even if you never use K3. 📩 Free newsletter: https://theaibridges.substack.com 🌐 Guides + coaching: https://theaibridges.com 00:00 Intro — a Chinese open model tops the coding leaderboard 00:51 Kimi K3 benchmarks vs Claude and GPT 02:53 Did it really pass Claude Opus? 05:30 Why is Kimi K3 priced like Claude? 07:48 The catch: hallucination rate and per-task cost 10:36 Did China just pass Claude? #KimiK3 #AINews #ClaudeAI