Note sull'episodio
Open source has caught up... again. A new model, Kimi K2 Thinking, has landed and is now ranked #2 globally—beating Grok 4, Claude 4.5 Sonnet, and Gemini 2.5 Pro. 🤯 This is the new "DeepSeek moment."
We’ll talk about:
- A deep dive into the stunning capabilities of the new open-source Kimi K2 Thinking model (Part 1).
- Real-World "One-Shot" Builds: We tasked it with building a drag-and-drop website builder (like Wix), a Photoshop clone with a working layer system, and an interactive 3D map of Tokyo.
- The Impossible Test: Kimi K2 successfully built a complex 3D fluid dynamics simulation—a test that Grok 4, Claude 4.5, and Gemini 2.5 Pro all failed.
Parole chiave
GPT-5Vibe CodingKimi K2Open Source AIAI Models