Google Cloud Previews AI Metadata Extraction in Cloud Storage

What happened
On November 19, 2026, Google Cloud introduced a preview service called object enrichments for its Cloud Storage platform. This service uses AI models to extract structured metadata from images and PDFs stored in Cloud Storage. The metadata is stored as object contexts and can be accessed for analytics or by AI agents, with integration for querying in BigQuery. The service is available through the Cloud Storage console and API, and supports asynchronous processing of eligible files.
Key details
- Supports five built-in AI models, including object detection, face detection, image labeling, and PDF classification.
- Pricing example: $700 to process 10 million images with one image model; object contexts cost $3 per million per month.
- PDF classification is free during the preview, with a cap of 100 million objects per customer.
- Supported image formats include BMP, GIF, ICO, JPEG, PNG, TIFF, and WEBP, with a 20 MiB size limit per object.
- Processing is available in the US, EU, and Asia regions, but not on Rapid or hierarchical-namespace buckets.
Why it matters
Enterprise customers using Google Cloud Storage can now automate metadata extraction from large volumes of images and PDFs, reducing manual labeling and custom pipeline development. This enables faster search, analytics, and AI training on stored assets, with cost savings compared to other providers. Data teams and AI developers can access structured metadata directly in BigQuery, streamlining workflows for asset discovery and multimodal analytics.
