Mux

Promptfree, for everyone, and the only version there is

Copy the prompt for Claude CodeCopy the prompt for CodexCopy the prompt for Cursor

Copy

Do not build this at scale. Transcoding and delivery at any scale is the product, and both are bills rather than code.

Uploading a file and receiving a stream means a ladder of renditions, packaging, storage and a delivery network. Every piece exists as open software; the cost is processor hours and egress, and at volume that is the whole conversation.

WHEN TO KEEP PAYING
- Video is central to the product and the volume is unpredictable
- You would rather not operate a transcode queue that falls behind on a busy day

WHEN TO BUILD
- A modest, predictable amount of video. Then ffmpeg on one machine and an object store with a delivery network in front is a fraction of the price, and the shape is under Wistia and Tella elsewhere in this catalogue

WHAT IT ACTUALLY INVOLVES
- A ladder of renditions — several heights and bitrates — packaged as HLS so a player adapts to the connection
- Transcoding queued with a concurrency cap. It is CPU-hours per video and unbounded parallelism takes the machine down
- The original kept untouched, always, so a re-encode never means re-uploading
- **A delivery network in front is not optional.** Serving video from one machine is the mistake that makes everything feel slow, and it is the part money genuinely solves
- Signed, expiring playback URLs where the video is not public, with the manifest and the segments both protected
- Captions from a local speech model, editable — a video without them excludes people and is unsearchable

MEASURE BEFORE DECIDING
Take your actual library: hours of video, average bitrate, and views per month. Multiply out the egress and price it against both options. Most people discover either that the hosted price is obviously reasonable, or that it obviously is not — and either answer is better than a preference.

AND THE FREE PART EVERYBODY SKIPS
Do not write a player. Use the browser's own video element with an HLS library, add captions and keyboard controls, and keep it under 40KB.

THE ONE-LINE VERSION
Work out your egress before anything else. The build is ffmpeg and a network; the decision is arithmetic.