Google Gemini, built as a multimodal model, can analyze PDFs, images, and video directly. Users can upload files into the prompt box to process visual elements, document layouts, charts, and raw media. For complex PDFs and documents, Gemini reads both text and visual structure, supporting uploads up to 100MB per file and up to 10 files per prompt. For screenshots and diagrams, its vision capabilities perform OCR and explain visual relationships. Video files up to 2GB per file and up to 5 minutes total length can be uploaded to extract insights, such as identifying mentions of a launch date or concerns about a timeline. For code repositories, users can import an entire folder or GitHub repository of up to 5,000 files per repo, with a total upload limit of 100MB. Audio files are limited to 100MB per file and up to 10 minutes total length.