Elon Musk’s xAI Unveils Grok 1.5 Vision AI Model in Preview, To Compete With GPT-4 Vision and Gemini Pro 1.5

Elon Musk's artificial intelligence (AI) firm xAI has unveiled a new AI model dubbed Grok 1.5 Vision. This large language model (LLM) is an enhanced version of the recently released Grok 1.5 model. With this upgrade, the AI model is now equipped with computer vision, making it capable of accepting visual media as input. It can process images and answer questions about it. Notably, the announcement came just days after OpenAI introduced its own computer vision-powered GPT-4 model.

The announcement was made by the official X (formerly known as Twitter) account of xAI. The firm shared a blog post detailing the new AI model and shared some of its benchmark scores. Since the vision capabilities were added to the recently unveiled Grok 1.5 model, most of the details remain the same. It has the same context window of 1,28,000 tokens and the general benchmark scores are also likely to remain the same.

xAI also shared benchmark scores of Grok 1.5 Vision tested on a benchmark developed by the company. The AI firm calls it the RealWorldQA benchmark and it measures “real-world spatial understanding”. It also tested the model in several other benchmarks such as MMMU, Mathvista, ChartQA, and more. While Grok outperformed OpenAI's GPT-4 with Vision and Gemini 1.5 Pro in RealWorldQA, it scored less in MMMU and ChartQA.

For the unversed, computer vision is a branch of computer science that deals with equipping computers (and AI models) with the ability to identify and understand objects in the real world using images and videos. This is designed to help computers see and process visual signals the way humans do. With the rise of multimodal AI models, many firms are now focusing on developing vision-focused models. Google's Gemini 1.5 Pro and OpenAI's GPT-4 with Vision both have this capability.

This technology also offers a wide range of applications. The Indian calorie tracking and nutrition feedback platform Healthify recently added a feature called Snap where users can click a picture of a food item or cuisine, and GPT-4 with Vision-powered AI chatbot suggests how the recipe can be made healthier, and how much exercise one needs to do to burn the extra calories. In future, AI models with computer vision can assist in the diagnosis of diseases, building self-driving cars, and more.