last highlighted date: 2024-06-12 Highlights Also, when considering Llama’s performance in English, non-code prompts, the 70b parameter model is second only to OpenAI’s GPT-4 Turbo (As per LMSYS leaderboards).