Reverse image search
You have a photo and want to know what else in your own library looks like it — not visually identical, just similar enough that a person would say 'oh, those go together.' This endpoint takes a reference image and a library of candidates and ranks the candidates by visual and conceptual resemblance, no captions, tags or file names required.
Run it online
Run this on our servers with your account. Free tools run in your browser; this one bills your KIT balance per the price above.
Why tags were never a real solution
For years the only practical way to search a large image library was to search the text around the images — file names, alt text, manually applied tags — which meant an image was only findable to the extent someone had bothered to describe it accurately. A mislabeled product photo, an untagged design asset, or a scanned image with no metadata became effectively invisible, not for lacking value but because the search system had nothing to read except text that was never guaranteed complete or correct.
What happens when you submit a reference image
POST a reference image along with the set of candidate images to compare it against to /search/similar-images and the task represents the visual content of every image — composition, subject matter, color, texture, and conceptual similarity — and ranks the candidates by how closely each one resembles the reference, all without touching any text metadata. The result is a list of matches ordered by resemblance, so a duplicate photo shot from a slightly different angle and a completely different image that happens to share the same subject and color palette can both be found correctly.
From pixel matching to meaning-aware comparison
Early reverse image search relied on comparing raw pixel patterns or simple perceptual hashes, which worked for near-identical copies but broke down the moment an image was cropped, recolored, or simply of a different but related subject. Representing images by their visual and conceptual content rather than raw pixels makes it possible to match two photos of similar products shot in different lighting, or two logos with the same shape but different colors — comparisons a hash-based system was never built to make.
Where it does real work
E-commerce catalogs use it to find duplicate or near-duplicate product listings across thousands of uploaded images; design and marketing teams use it to search an internal asset library by uploading an example instead of guessing what someone tagged it as years ago; stock and archive libraries use it to surface visually related images a keyword search would completely miss. In every case, the input a person actually has on hand is an image, not the words someone else once used to describe it.
Pricing and delivery
The endpoint bills a small base fee per request plus a per-query charge, published on this page regardless of how large the candidate library is. It processes asynchronously: submit the reference image and candidates, receive a task_id right away, and collect the ranked matches by signed webhook or a signed link valid for 24 hours; a failed task retries automatically up to three times and is never charged if it still can't complete.
What you can do with it
Duplicate product photo detection
An online marketplace scans newly uploaded listing photos against its existing catalog to flag likely duplicate or reused product images.
Design asset library search
A marketing team uploads a sample image to find visually similar assets in a large internal library instead of relying on inconsistent manual tagging.
Visual product discovery
A retailer lets shoppers upload a photo of an item they like and surfaces visually similar products from the catalog automatically.
Stock and archive image retrieval
A media archive finds visually related images across a large historical collection where captions are missing, inconsistent, or unreliable.
FAQ
What is a reverse image search API and how does it work?
It compares a reference image against a set of candidate images by visual and conceptual content, then ranks the candidates by how closely each resembles the reference, without relying on tags or captions.
Do my images need to be tagged or captioned first?
No, the comparison works directly on image content; no captions, alt text, or manual tags are required for the endpoint to find similar images.
Is the reverse image search API free to use?
There's no free tier — free tiers get abused and slow everyone down. Access runs on a prepaid ForHosting KIT balance: top up from $10.00 (it never expires) and each request is charged at its published price, so a call with no balance returns HTTP 402. No subscription, no tokens, no invented credits, and a failed task is never charged.
Can it find images that are similar but not identical?
Yes, it ranks by visual and conceptual resemblance rather than exact pixel matching, so cropped, recolored, or differently lit versions of a similar subject can still be found.
How do I search my own private image library?
Submit your reference image along with the candidate images from your own library in the same request; the task only compares the images you send, not any external image database.
How do I get the results back?
The task returns a task_id immediately and processes asynchronously, with the ranked matches delivered by signed webhook or a signed link valid for 24 hours.
Can I search a large image library at once?
Yes, submit the reference image with your full candidate set as a single request; pricing is per request and per query with no artificial cap on library size.
What happens if a reverse image search request fails?
The task retries automatically up to three times before returning a clear error, and a failed request is never charged.
For developers — API access
Everything on this page is available programmatically. This section is for teams who want to wire it into their own systems; everyone else can just use the tool above.
API endpoint
Prefer to automate it? One authenticated POST creates the task; the result comes back by webhook or a signed link. The same capability also runs here on the web, and soon from our app, email and Telegram.
Call it from your stack
curl -X POST https://api.kit.forhosting.com/search/similar-images \
-H "Authorization: Bearer $KIT_KEY" \
-H "Content-Type: application/json" \
-d '{"input":"…"}'const res = await fetch("https://api.kit.forhosting.com/search/similar-images", {
method: "POST",
headers: {
"Authorization": `Bearer ${process.env.KIT_KEY}`,
"Content-Type": "application/json"
},
body: JSON.stringify({
"input": "…"
})
});
const { task_id } = await res.json();import os, requests
res = requests.post(
"https://api.kit.forhosting.com/search/similar-images",
headers={"Authorization": f"Bearer {os.environ['KIT_KEY']}"},
json={
"input": "…"
},
)
task_id = res.json()["task_id"]<?php
$res = file_get_contents("https://api.kit.forhosting.com/search/similar-images", false, stream_context_create([
"http" => [
"method" => "POST",
"header" => "Authorization: Bearer " . getenv("KIT_KEY") . "\r\nContent-Type: application/json",
"content" => '{"input":"…"}',
],
]));
$task = json_decode($res, true);body := bytes.NewBufferString(`{"input":"…"}`)
req, _ := http.NewRequest("POST", "https://api.kit.forhosting.com/search/similar-images", body)
req.Header.Set("Authorization", "Bearer "+os.Getenv("KIT_KEY"))
req.Header.Set("Content-Type", "application/json")
res, _ := http.DefaultClient.Do(req)Example request
{
"input": "…"
}Example response
{
"task_id": "tsk_a1b2c3d4e5f6a1b2c3d4e5f6",
"type": "search.similar_images",
"status": "queued",
"_links": {
"result": "/tasks/tsk_…/result"
}
}The API is asynchronous: the call returns a task_id immediately and the result arrives by webhook. Polling is capped at 1 req/s per task.
Pricing
Published price — no tokens, no invented credits. A failed task is never charged.
Limits
max_chunks | 10000 |
max_tokens | 20000 |
Errors
| HTTP | Code | Meaning |
|---|---|---|
401 | unauthorized | Missing or invalid API key. |
402 | insufficient_balance | Your balance doesn't cover the task price. |
404 | unknown_type | That task type doesn't exist. |
429 | rate_limited | Too many requests. Use the webhook instead of polling. |