Grok, I wasn't familiar with your game.

xAI, Elon Musk's AI company, has launched Grok 1.5 Vision, a multimodal version of its Grok chatbot that can now process and understand visual information. The announcement, made via Musk's X platform, highlighted the model's strong performance on the RealWorldQA benchmark, where it achieved a leading score of 90.4%. This benchmark tests an AI's ability to answer questions about real-world spatial scenarios, such as interpreting traffic situations from a driver's perspective or understanding object relationships in a room. The release positions Grok as a more direct competitor to established multimodal models like OpenAI's GPT-4V and Google's Gemini, which have dominated the space.

The technical details reveal Grok 1.5V can analyze documents, diagrams, charts, screenshots, and photographs. xAI claims it matches or exceeds the capabilities of existing frontier models in several areas, including OCR (optical character recognition) and real-world spatial reasoning. The model is now available to early testers and premium subscribers on the X platform. This move signals xAI's rapid iteration pace, following the recent release of the text-based Grok 1.5, and intensifies the feature war among AI assistants, pushing the entire field toward more sophisticated, context-aware AI that can interact with the visual world.