Kimi K3 | First impressions

Try these yourself and see if you reach the same verdict: https://arena.ai/agent Kimi K3 just dropped from Moonshot, and the narrative on social media is that Chinese open models have caught up to the frontier. So we put it head-to-head against the leading models from OpenAI and Anthropic on the same prompts. We start with quick 3D and game generations for a visual sense check, then move into the tasks that matter for day-to-day work: research and data visualization. Check out the latest rankings on all the Arena leaderboards: https://arena.ai/leaderboard Peter’s prompts: https://github.com/petergpt/3d-prompt... 0:00 Intro: Kimi K3 and the "frontier-level" narrative 1:29 3D scene tests (Acropolis and more) 9:11 Game generation test 10:33 Why research + visualization tasks matter 11:31 Earthquake dataset visualization 15:26 AI model market research presentations 21:55 Quick SVG test 22:29 AI meeting assistant recommendation 27:18 Boeing 787 supply chain reconstruction 31:24 AI data center build-out research 35:26 Grocery basket price comparison 36:35 Auditing NVIDIA management claims 40:45 Swiss Alps hike planner 44:48 Verdict: price vs. performance #arenaai #moonshotai #kimik3 #opensource