🌐 LLM Leaderboard Update 🌐 LiveBench: Grok 4.20 Beta blasts in at #18 with 67.96, up from Grok 4's #20—GPT-5 Mini & DeepSeek slide back! New Results- === LiveBench Leaderboard === 1. GPT-5.4 Thinking xHigh Effort - 80.28 2. Gemini 3.1 Pro Preview High**5th rank in unseen questions across all categories - 79.93 3. Claude 4.6 Opus Thinking High Effort - 76.33 4. Claude 4.5 Opus Thinking High Effort - 75.96 5. Claude 4.6 Sonnet Thinking Medium Effort - 75.47 6. GPT-5.2 High - 74.84 7. GPT-5.2 Codex - 74.30 8. GPT-5.1 Codex Max High - 73.98 9. Gemini 3 Pro Preview High - 73.39 10. GPT-5.3 Codex High - 72.76 11. Gemini 3 Flash Preview High - 72.40 12. GPT-5.1 High - 72.04 13. GPT-5 Pro - 70.48 14. Kimi K2.5 Thinking - 69.07 15. GLM 5 - 68.85 16. GPT-5.1 Codex - 68.61 17. Claude Sonnet 4.5 Thinking - 68.19 18. Grok 4.20 Beta - 67.96 19. GPT-5 Mini High - 65.91 20. DeepSeek V3.2 Thinking - 62.20 #ai #LLM #LiveBench