Google has officially launched Gemini, its latest artificial intelligence model, asserting that it surpasses OpenAI’s GPT-4. Optimized for various sizes and use cases (Ultra, Pro, Nano), Gemini is a multimodal model designed to comprehend and combine different types of information. Notably, it outshines GPT-4 in advanced math and specialized coding capabilities.
Google claims Gemini Ultra achieves “state-of-the-art performance” across 30 out of 32 academic benchmarks, scoring 90% on a massive multitask language understanding (MMLU) test, surpassing human expert performance.
Gemini’s multimodal capabilities, including reasoning across text, images, audio, and video, position it ahead of competitors. It also boasts advanced programming skills, generating high-quality code using AlphaCode 2. Google has fine-tuned Gemini Pro for its ChatGPT counterpart, Bard, making it “the most powerful free chatbot on the market today.” The AI model is set to roll out on Google’s Pixel 8 Pro and will be deployed across various Google products and services in the coming months, promising a generative search experience.
