Latest #AI benchmark results: DeepSeek-R1 (including its distilled variants) outperforms OpenAI's o1-mini and preview models. And the Llama 3 distilled version now holds the title of the highest-performing LLM I've tested locally to date. 🚀