Pricing

One price per file type, charged per minute of video or audio and per image. Every job runs the full analysis. There are no tiers and no subscription. New accounts start with $5 of credit.

20¢
per minute of video
video/footage, video/edit
Minimum 20¢ per job
per minute of speech
audio/speech
Minimum 5¢ per job
per minute of music
audio/music
Minimum 2¢ per job
per image
image/scene, image/graphic
Minimum 3¢ per job

A 10-minute video costs $2. One thousand photos cost $30. The $5 starting credit covers 25 minutes of video.

Credit

You buy credit in packs on the platform, in $20, $100, and $500 packs or any whole amount from $10 to $5,000. Credit is prepaid and doesn't expire. When the API accepts a job, it holds the estimated price. When the job ends, it charges the actual cost or releases the hold.

Limits

A file can be up to 10 minutes long, and an upload can be up to 300 MB. The analysis only needs a 720p copy. For a larger video, upload a proxy and describe the original in the request; the API then writes the result about the original. Each key allows 60 requests per minute by default; you can change that per key on the platform.

What every job includes

Video

ffmpeg reads the file, detects the cuts, and measures the loudness. OpenCV measures the camera movement. Pixel statistics give the color and tone. A video model describes the actions, the people, and what's on the frame. Deepgram times the words and separates the speakers. librosa and ACRCloud analyze the music.

Audio

The samples give the loudness, noise floor, dynamic range, clipping, and a waveform. ffmpeg finds the silences. Deepgram produces the words, sentences, paragraphs, and speakers. A language model describes the voices and writes the summary. librosa finds the tempo, key, and beats. ACRCloud identifies the recording.

Images

The file's metadata gives the camera, lens, exposure, capture time, and GPS. Pixel statistics give the palette, tone, and sharpness. A vision model identifies the kind of image, its form and composition, the people and objects with their bounding boxes, the text, and the surroundings.

What each of the six types returns, field by field, is in the docs.