Google’s ‘GPT-4 killer’ Gemini is out, here’s how you can try it

Google has deployed its newest weapon in the AI arms race, a new artificial intelligence model that it claims is smarter and more powerful than OpenAI’s GPT-4.

Tech behemoth Google has officially introduced Gemini, its latest artificial intelligence model, which it asserts surpasses OpenAI’s GPT-4 in capabilities.

On December 6, Google’s CEO Sundar Pichai and Demis Hassabis, CEO and co-founder of Google DeepMind, announced Gemini’s launch through a company blog post. This AI model is designed to be versatile, with different versions tailored for various needs (Ultra, Pro, Nano), and is multimodal, enabling it to understand and integrate different types of data. Notably, Gemini boasts superior abilities in mathematics and specialized coding compared to OpenAI’s GPT-4, which has limitations in performing mathematical tasks.

Google highlights that its Ultra version of Gemini achieves top-tier performance in 30 out of 32 academic benchmarks commonly used in the development of large language models (LLMs). Additionally, it excels in the massive multitask language understanding (MMLU) test, scoring 90%, a feat that surpasses human expert levels, as per Google’s claims.

Jeff Dean, Google’s chief scientist, emphasized that Gemini Ultra is the first model to reach human-expert performance on MMLU across 57 subjects, achieving a score above 90%. The model is uniquely engineered to seamlessly process and reason across various formats, including text, images, audio, and video, positioning it ahead of its competitors.

Dean explained, “Gemini was conceived as a multimodal model from the beginning, unlike starting with a text-only model and later adding vision and audio capabilities.” Furthermore, Gemini showcases advanced programming skills, capable of generating high-quality code through AlphaCode 2, an advanced code generation system. It can solve intricate programming challenges and assist developers in their work.

AI specialist Rowan Cheung noted that Gemini Pro outperformed GPT-3.5 in six out of eight benchmarks, making it the most potent free chatbot available. Google has integrated a refined version of Gemini Pro into its ChatGPT equivalent, Bard, offering users an enhanced experience.

“This is Bard’s most significant update since its debut. It’s now accessible in English in over 170 countries and territories, with plans to expand to more modalities, languages, and regions soon,” Google stated.

Additionally, Gemini is being incorporated into Google’s latest smartphone, the Pixel 8 Pro. “The Pixel 8 Pro is the first to utilize Gemini Nano, powering new features like the Summarize option in the Recorder app and soon to be featured in Smart Reply in Gboard for WhatsApp, with more messaging apps to follow next year,” the company announced.

Google plans to integrate Gemini into more of its products and services, including Search, Ads, and Chrome, in the upcoming months. The company is also exploring the use of Gemini to enhance its dominant web search engine, aiming to transform searching into a generative experience.

Google had previously unveiled Gemini earlier in the year, touting its advanced capabilities and suggesting it would outperform ChatGPT.