Developer Infrastructure Comparison

Xora vs. ffmpeg-api.com

Both run real FFmpeg in the cloud, with filter graphs, async jobs and an MCP server. The big difference is the meter — ffmpeg-api.com multiplies file size by processing time. Here is what that does to your bill.

Featureffmpeg-api.comXora (xora.sh)
Billing ModelGB-seconds: (input GB + output GB) × processing seconds Credits — processing time + output produced
Entry Paid Plan$11 / mo for 500 GB-seconds$15 / mo for 400 credits
Max Input File500 MB Starter · 1 GB Pro · 5 GB Growth ($99) 250 MB free · 2 GB Build ($15) · 6 GB Scale ($49)
Custom FFmpegInput options, output options, maps and a filter_complex stringOne args array — the command you run locally
Pre-built RecipesNo (plain-English AI endpoints in beta instead) Yes (9 recipes like webReady, plus a probe mode)
Storage ArchitectureOutputs saved to ffmpeg-api.com directories, returned as download links Managed storage with signed URLs, or your own Cloudflare R2 bucket (BYOB, paid plans)
AI Agent AccessRemote MCP server that runs jobs (OAuth or API key)Remote MCP server that runs jobs (OAuth or API key), plus /llms.txt and OpenAPI
Free TierFree plan; limits not published 25 credits every month (no card required)

Billing Mechanics: GB-Seconds vs. Credits

A GB-second bill grows with file size and processing time at the same time, so a bigger file costs more twice. Xora charges for the processing time and the output separately, and adds them.

Scenario A

10-min 1080p transcode

A full re-encode, 1.5 GB in and 0.8 GB out, with about 4 minutes of processing.

ffmpeg-api.com Pricing

~$10.90

2.3 GB × 240 s = 552 GB-seconds. The 1.5 GB file needs the Growth plan ($99 / 5,000 GB-s), and this one job uses 11% of it.

Xora Pricing

~12 credits

About 4.3 minutes of processing + 8 credits for the 0.8 GB output — about $0.45 at Build's ~4¢/credit.

Xora is ~95% cheaper once files are bigger than a few hundred MB

Scenario B

Quick edit of a 20 MB clip

A short social clip, 20 MB in and 10 MB out, with about 5 seconds of processing.

ffmpeg-api.com Pricing

~$0.003

0.03 GB × 5 s = 0.15 GB-seconds, at Starter's $11 / 500 GB-s.

Xora Pricing

~0.2 credits

5 seconds of processing + 0.1 credits for the 10 MB output — about $0.007 at Build's ~4¢/credit.

GB-seconds can win on tiny, fast files — credits win as soon as files grow

Split Task Fields vs. One Args Array

ffmpeg-api.com supports real FFmpeg work, including filter graphs. To send a command, you split it into its parts: options for each input, a separate filter_complex string, then options and stream maps for each output.

Xora takes the arguments array in the same order you run it locally. Test the command on your laptop, put each argument in the array, and send it. Your agent can do the same with no new schema to learn. Same args, different machine.

ffmpeg-api.com payload (split task fields)
{
"task": {
"inputs": [
{ "file_path": "https://example.com/source.mov" },
{ "file_path": "https://example.com/logo.png" }
],
"filter_complex": "[0:v][1:v]overlay=W-w-10:H-h-10[out]",
"outputs": [{
"file": "output.mp4",
"options": ["-c:v", "libx264", "-crf", "23"],
"maps": ["[out]", "0:a"]
}]
}
}
Xora payload (one args array)
{
"mode": "ffmpeg",
"input_files": {
"in_1": "https://example.com/source.mov",
"in_2": "https://example.com/logo.png"
},
"output_files": { "out_1": "output.mp4" },
"ffmpeg": {
"args": [
"-i", "{{in_1}}", "-i", "{{in_2}}",
"-filter_complex", "[0:v][1:v]overlay=W-w-10:H-h-10[out]",
"-map", "[out]", "-map", "0:a",
"-c:v", "libx264", "-crf", "23",
"{{out_1}}"
]
}
}

Two Major Architecture Upgrades

1. A Bill You Can Predict

With GB-seconds you must know the file size and the processing time before you know the cost, and the processing time depends on their machines. A file twice as big that takes twice as long costs four times as much.

Xora meters processing time and output size, and adds them. With stageInput: false, a thumbnail from a 2 GB file is about 0.1 credits because a ranged read takes only the bytes it needs. Every job reports its own cost, and jobs that fail, are rejected or are cancelled cost nothing.

2. Direct Storage Delivery (BYOB)

ffmpeg-api.com saves outputs to its own directories and gives you download links. Its docs do not say how long files stay there.

On paid plans, Xora can write finished files straight into your own Cloudflare R2 bucket. On any plan, managed storage gives you signed URLs and a Files API to list and delete outputs, so you always know where your media is.

See all options in the best FFmpeg APIs compared, and Xora’s pricing.

Ready to try Xora?

Bring the FFmpeg command you already have, run it with one POST, and pay only for the credits your jobs actually use.