Google's new Gemini AI model tops LLM leaderboard in rapid debut

Google has launched a new version of its Gemini AI model, named Gemini-Exp-1114, which has quickly reached the top of the Imarena Chatbot Arena leaderboard. This model surpassed OpenAI's latest GPT-4o and the o1-preview reasoning model. Gemini-Exp-1114 excels in math and vision tasks, areas where previous Gemini models have also performed well. Currently, it is not available on the Gemini app or website but can be accessed through a free Google AI Studio account. The new model's success coincides with the release of a Gemini app for iPhone, which has outperformed the ChatGPT app in recent comparisons. The top five models on the leaderboard are all from Google or OpenAI, with xAI's Grok 2 being the first non-Google/OpenAI model.