Open PicVault

What PicVault cannot do

Checked against the running server on 2026-10-28

Photo downloaders usually tell you what they can do and stay quiet about the rest. This page is the rest, with the actual figures, so a failure reads as a known boundary instead of a mystery.

The short version

  • Too fast. Six link readings and thirty photo downloads per address per minute. Separate counters.
  • Too big. 150 MB per package, 25 MB per photo, 30 slides per selection.
  • Too many at once. Two ZIP builds run together; the rest wait rather than fail.
  • Too slow to use. A selection expires fifteen minutes after it is made.
  • Not accepted. Profile links, private posts, Stories and video-only slides.
The fifteen-minute clock on a selection is the one that catches people out. It starts the moment you make the selection, not when the ZIP finishes — so a selection left open in a tab overnight is gone in the morning, and the error you get says expired rather than not found.

The thirteen limits, with the numbers

1. Six link readings per minute from one address

Reading a post link — the step that lists the slides — is capped at six per minute per client address. The seventh in the same minute is refused with HTTP 429 and a retry delay of sixty seconds. The counter is a fixed one-minute window, not a rolling average, so it resets all at once rather than trickling back.

2. Thirty photo downloads per minute from one address

Saving individual photos is capped at thirty per minute per address, which is generous on purpose: a large carousel is meant to be saved photo by photo. This counter is separate from the link-reading one, so being blocked from reading does not block you from downloading what you already listed.

3. Two ZIP files build at the same time

Packaging is capped at two concurrent builds. The third waits rather than failing. This was raised from one after measurement: with a single slot, one slow image host stalled every other packaging request behind it, and the waiting page showed nothing for forty seconds or more. Two slots keep one slow source from blocking everyone else.

4. A forty-post selection in memory is not a saved list

A selection lives in server memory for fifteen minutes from the moment it is made. Past that the identifier expires and the download link stops working, and the answer to an old link is an expired-selection error rather than a stale file. Fifteen minutes is enough to click through a selection and start the ZIP; it is not long enough to treat as storage.

5. A package cannot exceed 150 MB

When the running total of the archive passes 150 MB the build aborts. The cap applies to the sum, not to any single image, so the practical limit depends on how large the photos in that post happen to be. A post of unusually large originals can hit it with fewer photos than you would expect, and the answer says the package was too large rather than quietly truncating the file.

6. A single photo cannot exceed 25 MB

Any one image larger than 25 MB is refused during transfer, before it is written to the archive. On Instagram this is rare for photos — 25 MB is far above a normal upload — so in practice this is a guard against a hostile or malformed source rather than something a real post will trip.

7. Thirty slides is the most one post can contribute

A selection is bounded at thirty entries. A carousel with more slides than that cannot be taken in one pass; the extra slides are not silently dropped from the middle, the selection itself is rejected as out of bounds. Thirty is also above what Instagram normally allows in one post, so this mostly matters for posts assembled from mixed media.

8. Video slides are skipped, and the numbering shows it

Carousels can mix photos and videos. A video slide is not extracted and does not occupy a photo file name, so the surviving photos keep the position numbers they had in the original post and the sequence has gaps. This is deliberate: renumbering them from one would make it impossible to match a saved file back to a slide.

9. Only post links are accepted, never profile links

The tool reads a post address carrying a shortcode. A profile address is refused outright — there is no browsing of an account, no listing of its recent posts and no batch saving of a whole feed. Stories and Reels are outside this tool as well: Reels belong to a separate video tool, and Stories are not reachable at all because they are not public in the way a post is.

10. Private posts are unreachable

Nothing here logs in, so nothing behind a login can be read. A private account, a followers-only post, or a post that has been deleted returns the same not-found answer. There is no way around this from the page, and no setting that changes it.

11. Quality is whatever the source stored, never improved

The tool saves the largest version of each image that Instagram serves, and nothing more. There is no upscaling, no sharpening and no re-encoding. A photo uploaded small stays small, and an old post may resolve to a smaller version than a recent one because of what the source kept, not because of anything this tool did.

12. Two hundred selections exist server-wide at once

Selections are held for everyone on the server together, with room for two hundred. Beyond that the oldest are dropped to make space, which means a very quiet selection can expire earlier than fifteen minutes under heavy load. Individual photo downloads are unaffected by this ceiling — only the packaged ZIP path uses it.

13. Which failure codes can be retried, and which cannot

RATE_LIMITED and TIMEOUT are worth retrying; QUOTA_EXCEEDED is worth retrying only after selecting fewer slides. INVALID_URL, NOT_FOUND_OR_PRIVATE, VIDEO_ONLY, EXPIRED, NOT_FOUND and UPSTREAM_ERROR are permanent for that request, and retrying them changes nothing. All ten are listed under each failure above, in the same words the API returns.

What each failure actually means

Seven outcomes look alike from the outside. Only three of them are worth reacting to, and none of them mean the link was wrong.

1Rate limitedYou used this minute’s allowance. The answer carries a sixty-second retry delay; waiting it out is the entire fix. RATE_LIMITED
2Quota exceededThe selection is out of bounds — usually more entries than one post can contribute. Select fewer slides. QUOTA_EXCEEDED
3ExpiredThe selection aged out, or the identifier was already cleaned up. Make the selection again; nothing else is wrong. EXPIRED
4Not found or privateThe source itself will not serve the post. Retrying changes nothing. NOT_FOUND_OR_PRIVATE
5Video onlyThe link points at a video post with no photos in it. This is a photo tool; the answer is a different tool, not a retry. VIDEO_ONLY
6TimeoutThe source did not answer in time. Worth one retry, especially on an old post. TIMEOUT
7Upstream errorThe source answered with something unusable. Usually transient, occasionally permanent for that post. UPSTREAM_ERROR

Three more codes exist and are not failures of a post: INVALID_URL means the link is not a post address at all — a profile, a Story, a Reel or another host — and no retry will fix it; NOT_FOUND means the identifier is unknown to the server, which happens to a selection that was never made here; NETWORK means the request never reached the server, so the answer says nothing about the post.

If you are automating this: the rate limit answers with HTTP 429 and a retry delay, while a size or selection problem answers with HTTP 413. Treating both as "try again later" is how an automated client turns a permanent rejection into an endless retry loop that also burns the per-minute allowance.

What is not stored

No account, so there is no library, no history and nothing to sign back into. A selection lives in server memory for fifteen minutes and is then dropped. Photos pass through and are not kept as a collection. There is no record of which posts have been downloaded by whom, because there is no "who" in the first place.

What this tool is not

Not affiliated with Instagram or Meta

PicVault is independent. It is not affiliated with, sponsored by or endorsed by Instagram or Meta, and a refusal or a failure says nothing about what those platforms would allow.

Not a licence to republish

A public post is not permission to reuse someone else’s photograph. Public visibility and copyright are separate questions, and saving a file does not answer the second one.

Not an archiver

There is no way to point this at an account and collect everything it has ever posted. One post at a time, by link. That boundary is deliberate, not a missing feature.

PicVault

Paste a public Instagram post link, pick the photos, save them one by one or as a ZIP. No app, no Instagram login.

Open PicVault

More from PicVault

Instagram carousel downloader Download Instagram photos from a link Instagram photos in the highest available quality Is it legal to download Instagram photos? Save Instagram photos to a computer