Fetch: download media from a URL

Verinio Fetch is an API endpoint that downloads media from a supported source URL and returns the file, or lists the available formats and qualities first. It replaces running your own scraper or yt-dlp instance: send a URL, get back a direct, working download. Previewing available formats has its own flat cost (see how pricing works), and downloads are priced by size (about 3.5 credits per MB, 5-credit minimum) since bandwidth — not the request itself — is the real cost. See Fetch vs. self-hosting →

Transcript: extract captions as text

Verinio Transcript pulls a YouTube video's transcript and returns it as TXT, SRT, VTT, or JSON — whichever format your pipeline or video editor expects. It's built for feeding video content into summarizers, search indexes, or subtitle files without scraping the page yourself. Both looking up transcript availability and downloading the transcript have a flat cost each; there's no file-size pricing here because transcript text is small. See Transcript vs. alternatives →

Metadata: inspect, sanitize, or compress a file

Verinio Metadata reads the embedded metadata of an uploaded file (EXIF, codec info, duration, GPS, device details), strips identifying metadata for privacy ("sanitize"), or compresses the file while keeping it usable. This is the one service you run on your own files rather than a source URL. Each action (inspect, sanitize, compress) has its own flat cost. See Metadata vs. exiftool/ffprobe scripts →

How they fit together

A typical agent pipeline uses all three in sequence: Fetch pulls a video down, Transcript extracts its captions for a summary or search index, and Metadata strips identifying details or compresses the file before storage or re-upload. Each service is a separate endpoint with its own credit cost, so you only pay for the steps you actually use. See the AI agent integration guide for how this looks as real API calls, or the API overview for the endpoint list.