ck:ai-multimodal

Long-form transcription and speaker identification use the Gemini API with cost-aware chunking to generate timestamped outputs. It handles audio analysis directly and writes the resulting text to your project assets directories.

Yes, multimodal media generation supports Gemini, Imagen, and MiniMax for images, plus Veo and MiniMax Hailuo for video. It includes model selection and billing fallbacks to manage generation costs across different providers.

Vision and OCR tasks process screenshots, product photos, and documents using Gemini for image understanding. It performs object detection, segmentation, and captioning, then writes the extracted text to project assets.

You need a GEMINI_API_KEY and required Python dependencies validated via the setup checker. The system supports API key rotation and model fallbacks to ensure continuous operation during multimodal media processing.

Media preflight optimization processes large files by supporting both inline uploads and the File API. This ensures efficient file upload handling before executing vision understanding, transcription, or generation tasks.

API key rotation prevents rate limiting during intensive video creation and TTS or music synthesis workflows. It automatically switches keys and performs model fallbacks to maintain uninterrupted multimodal media generation.