Getting started
Instagram transcript API
A single endpoint that takes a public Instagram link and returns the spoken words with timings. Sign in, buy credits, call it.
The API does one job. You send a public Instagram link. It sends back the speech as text, with a timestamp on every line, plus subtitle files.
It is the same pipeline behind the free tool on this site. The difference is that you call it from your own code, you get a much higher rate limit, and you can run it across many links at once.
Who this is for
Agencies and research teams pulling transcripts across a set of accounts on a schedule.
Content tools that let their own users paste an Instagram link.
Agents. If you are building with an assistant that needs to read short video, there is an MCP server that wraps the same endpoint.
Getting a key
Create an account, then make a key on your account page. It takes about a minute and nobody has to approve anything.
New accounts start with 10 free credits, which is enough to wire the endpoint up and see real output before you pay for anything.
What it costs
One credit is one minute of audio, rounded up. A 20 second Reel costs one credit. Packs start at $10 for 250 credits and they do not expire. The pricing page has the rest.
Credits come off after the transcript exists, so a private post or a timeout costs nothing.
What you get back
Every successful call returns the same object.
transcriptis the whole thing as one block of text.segmentsis an array of lines, each with a start and end time in seconds.srtandvttare ready-made subtitle files as strings.timestampedis a human-readable version with[mm:ss]in front of each line.languageis the detected language as a two-letter code.
You almost always want transcript or srt. The rest are there so you do not have to build them yourself.
What it will not do
Private posts. The endpoint reads what is publicly readable. A post behind a login returns PRIVATE. No key changes that.
Stories. Reels, ordinary video posts and IGTV links only.
Long video. There is a duration cap. Three minutes for a guest on the free tool, ten minutes on a key.
Audio. The video file lives in a temporary folder for the length of the request and is deleted when it ends. The words of a public Reel are kept for 30 days, keyed on the post, so a later call for the same link does not run the speech model again. We do not keep who asked.
Rate limits
Every response carries the usual headers.
| Header | Meaning |
|---|---|
x-ratelimit-limit |
Requests allowed in the current window |
x-ratelimit-remaining |
Requests left |
retry-after |
Seconds to wait, sent with a 429 |
Back off when you see a 429. Retrying immediately will not work and it counts against you.
Next
Start with the quickstart. It is one request.