Cap opens Cap Infra, its video and speech API, to developers

Cap opened the infrastructure behind its own recorder to everyone as Cap Infra, a video hosting and speech-to-text API.

Upload a video and you get adaptive HLS in up to five renditions from 360p to 4K, never above the source, plus one MP4, a poster image, a hosted player page and an iframe embed. To start, you create an app in the dashboard, copy its secret key and add credit. Your server creates an upload and gets a short-lived token, and the browser sends the file in parts straight to storage. A video counts as ready as soon as its first output can play, and the rest finish in the background.

Speech to text also works by itself. Send a public URL or upload audio or video, and you get text back with word and segment timings. Uploaded videos aren't transcribed automatically. There's a hosted MCP server too, and a setup prompt you can paste into Claude Code, Codex or Cursor.

Pricing is prepaid credit per app, with no seats or plans. Uploading, encoding and the player are free. Storage costs $0.85 per 1,000 minutes a month up to 1080p, $1.45 at 1440p and $2.80 at 4K. Delivery is $0.95 per 1,000 delivered minutes, billed from November 14. Speech to text is $0.10 per audio hour, and Cap Pro includes $5 of credit a month. Uploads are capped at 36 GB.

Cap says an hour of 4K is playable in under 5 seconds after the upload finishes and fully processed in about 2 minutes, and that an hour of audio is transcribed in under 24 seconds. Those figures come from its own tests. The API, media server and player are open source.