Google's Gemini 1.5 Pro Tops AI Leaderboard, Surpasses GPT-4o by 14 Points with 1300 ELO

Aug 1, 2024, 04:41 PM

Google has launched a major upgrade to its Gemini chatbot with the introduction of Gemini 1.5 Flash, a lighter and faster model that will power the free tier of Gemini on both web and mobile applications. Concurrently, Google DeepMind's experimental Gemini 1.5 Pro (0801) has been tested in the Chatbot Arena, gathering over 12,000 community votes and claiming the number one spot on the LMSYS leaderboard, surpassing GPT-4o and Claude-3.5 by 14 points. The new version is available for early testing on Google AI Studio and the Gemini API. This significant achievement has been attributed to various advances in post-training, and it marks a notable milestone for Google's AI capabilities, with an ELO rating of 1300.