ai-multimodal

Process and generate multimedia content using the Google Gemini API for enhanced vision capabilities. Features include analyzing audio files (transcription with timestamps, summarization, speech understanding, music/sound analysis up to 9.5 hours), understanding images (more advanced image analysis than Claude models, captioning, reasoning, object detection, design extraction, OCR, visual Q&A, segmentation, multi-image handling), processing videos (scene detection, Q&A, temporal analysis, YouTube URLs, up to 6 hours), extracting from documents (PDF tables, forms, charts, diagrams, multi-page), generating images (text-to-image with Imagen 4, editing, composition, refinement), and generating videos (text-to-video with Veo 3, 8-second clips with native audio). Use this when working with audio/video files, analyzing images or screenshots (instead of Claude's default vision capabilities; fall back to Claude's vision only if needed), processing PDF documents, extracting structured data from media, creating images/videos from text prompts, or implementing multimodal AI features. Supports Gemini 3/2.5, Imagen 4, and Veo 3 models with context windows up to 2M tokens.