Bing Images Scraper

A Bing Image Search API alternative. Type a keyword, get hundreds of unique full-size image URLs back with source pages and metadata, as JSON, CSV, or Excel. No Microsoft API key, no Azure subscription, no quota forms.

Microsoft retired the Bing Search APIs in August 2025; old v7 keys now return HTTP 410 and the official replacement, Grounding with Bing Search, needs Azure billing and restricts how you store results. This actor fills the gap and reaches further than the old API did: it queries Bing across multiple markets and proxy regions in parallel, so one keyword returns 200 to 350 unique images instead of the ~35 a browser shows.

📋 How to scrape Bing Images

1. Enter keywords in Search queries, one per line.
2. Toggle regions: Europe and US are on by default, Asia is opt-in.
3. Click Start. A query takes 10 to 20 seconds.
4. Open the Output tab and click Export for JSON, CSV, Excel, or HTML.

🎁 So what do you get?

🖼️ Full-size image URL	🔗 Source page	🧬 md5 hash for dedup
🏷️ Title and description	🗺️ Which Bing markets returned it	🌍 Which regions surfaced it
⚡ Bing CDN thumbnail	📌 Bing image and content IDs	🧾 Full provenance per row

⚖️ Compared to the retired Bing Image Search API

	Bing API v7	This actor
🚦 Status	❌ Retired, keys return HTTP 410	✅ Working
🖼️ Results per query	35 per page	✅ 200 to 350 unique
🔑 Access	❌ Azure key, quota tiers	✅ Register on Apify, $5 free monthly credit
📦 Output rights	⚠️ Storage restrictions in the successor	✅ Keep, cache, train on it
🧬 Dedup hash	❌	✅ md5 on every row

Migrating from Bing API v7

Porting code that parsed the old images/search response? The fields map one to one:

Bing API v7 (value[] field)	This actor
contentUrl	image_url
hostPageUrl	page_url
thumbnailUrl	thumbnail_url
name	title
imageId	mid
no v7 equivalent	md5, regions, markets, provenance

Not carried over: width, height, contentSize, and accentColor. If your pipeline depends on any of these, open a ticket on the Issues tab and I'll add them.

🎯 Three things people run this for

	How
🧠 ML training datasets	One query per class label; the query field on every row becomes the label. The fast.ai dataset workflow, without the dead API key
🛍️ Product and competitor research	Batch a list of product keywords and export one spreadsheet of images, sources, and titles
📈 SEO across regions	markets shows which Bing market surfaced each image. Searching cat returns very different images from Bing Japan than Bing Europe

📥 Input

{
  "queries": ["red running shoes", "blue running shoes"]
}

• queries: one search per keyword
• scanEU: default true. Bing's European markets (nl-BE, en-GB)
• scanUS: default true. Bing's US market (en-US)
• scanAsia: default false. Bing's Asian markets (ja-JP), worth enabling for Asian subjects
• marketOverride: a single Bing mkt code like pt-BR or ar-SA. Overrides the region toggles
• safeSearch: default false for maximum coverage. Set true for Bing's moderate filter

🧠 A dataset with one query per class

{"queries": ["running shoes", "hiking boots", "sandals"]}

🇧🇷 One specific market

{"queries": ["tênis de corrida"], "marketOverride": "pt-BR"}

📤 Output

One row per unique image, deduplicated by Bing's md5 hash within each query.

{
  "query": "bart de wever",
  "image_url": "https://upload.wikimedia.org/.../Bart_De_Wever_2025.jpg",
  "page_url": "https://nl.wikipedia.org/wiki/Bart_De_Wever",
  "thumbnail_url": "https://ts4.mm.bing.net/th?id=OIP...",
  "title": "Bart De Wever - Wikipedia",
  "md5": "138a8fac7b9e922fb529b9ce1ef932ae",
  "desc": "Belgian Prime Minister",
  "regions": ["EU", "US"],
  "markets": ["nl-BE", "en-GB", "en-US"],
  "proxy_countries": ["DE", "AE"]
  // HIDDEN: mid, cid, provenance (the full market/region/proxy tuple per sighting)
}

⚠️ Dedup is per query. The same image matched by two of your queries gives two rows, one per query, so billing matches exactly what you asked for. Group by md5 to dedup across queries yourself.

Fields

Field	Description
query	The search term that returned this image, your class label in dataset work
image_url	The full-size image on the source site, not a Bing thumbnail
page_url	The webpage hosting the image
thumbnail_url	Bing's CDN thumbnail, fast to display and virtually never blocked
title, desc	Image title and description, when Bing provides them
md5	Bing's content hash. The dedup key
mid, cid	Bing's internal image and content IDs
regions, markets, proxy_countries	Where this image was surfaced
provenance	Full (market, region, proxy_country) tuples, for debugging surprising results

🧠 Build an ML image dataset

The old fast.ai-style dataset loop died with the Bing API in August 2025. This is the same loop without the Azure key: one query per class label, run, then download from the URL list.

import pathlib
from curl_cffi import requests

for item in dataset_items:
    folder = pathlib.Path("dataset") / item["query"].replace(" ", "_")
    folder.mkdir(parents=True, exist_ok=True)
    try:
        r = requests.get(item["image_url"], impersonate="chrome", timeout=10)
        (folder / f"{item['md5']}.jpg").write_bytes(r.content)
    except requests.RequestException:
        pass  # dead source links happen; skip and keep the dataset clean

The folder-per-label layout drops straight into PyTorch's ImageFolder, Keras' image_dataset_from_directory, or fast.ai's ImageDataLoaders.

💡 Tip: the download uses curl_cffi instead of plain requests because some hosts refuse requests whose TLS fingerprint doesn't look like a browser. If a host still refuses, fall back to thumbnail_url, served from Bing's own CDN.

⚙️ Use it as a Bing Image Search API

Every run is an HTTP endpoint: POST the same JSON as the form and the images come back in the response body.

Python

import requests

resp = requests.post(
    "https://api.apify.com/v2/acts/thodor~bing-images/run-sync-get-dataset-items",
    params={"token": "YOUR_APIFY_TOKEN"},
    json={"queries": ["red running shoes"]},
)

for item in resp.json():
    print(item["image_url"], item["page_url"])

Node.js

import axios from "axios";

const { data } = await axios.post(
  "https://api.apify.com/v2/acts/thodor~bing-images/run-sync-get-dataset-items",
  { queries: ["red running shoes"] },
  { params: { token: process.env.APIFY_TOKEN } }
);

console.log(data.length, data[0].image_url);

curl

curl -X POST "https://api.apify.com/v2/acts/thodor~bing-images/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"queries":["red running shoes"]}'

Swap run-sync-get-dataset-items for runs to fire async with a webhook on big keyword lists. The apify-client SDK works too, in Python and JavaScript, and the n8n, Make, and Zapier integrations take the same input.

💡 Tip: no need to write the JSON by hand. Fill in the form on the Input tab, switch the editor from Form to JSON, and copy the result into your code.

💰 How much does it cost to scrape Bing images?

Billing is per result, at the rate on the price card on this page. No compute charges, no subscription, no minimum spend. A default run returns roughly 200 to 350 images per query and the Asia toggle adds about 50 more, so the region toggles are also your cost control: disable what you don't need.

❓ FAQ

Is the Bing Image Search API still available? No. Microsoft retired the Bing Search APIs in August 2025, and the replacement inside Azure AI Foundry requires an Azure subscription and restricts how you store results. This actor is the working alternative.

How do I get a Bing Image Search API key? You can't anymore; existing v7 keys return HTTP 410 Gone. If a tutorial or course asks for one, this actor is the drop-in replacement with no key at all.

Can I use this Bing images scraper for free? Yes. Registering on Apify comes with $5 of free platform credit every month, no credit card needed, enough for real test batches.

Why do some images look like duplicates? Bing's regional indexes rank overlapping images differently. Rows are deduplicated by md5 within each query, so lookalikes in the output are genuinely different files: cropped, resized, or hosted elsewhere.

Does this do reverse image search? Not yet, keyword search only. For finding where an image appears online, use the Reverse Image Search API. If you need Bing Visual Search specifically, open a ticket and I'll prioritize it.

Will Bing block me? Not under normal use. Requests go through residential proxies with browser TLS impersonation, and Bing is far more permissive than Google: no rate limits or captchas seen across thousands of test queries.

Can I use the images commercially? The actor returns URLs and metadata, not licences. Check terms per source via page_url. Nothing here grants rights to the underlying files.

🛟 Support

Something not working, or missing a feature like Bing Visual Search or a specific region? Message me in the Issues tab and I'll look into it quickly. I'm a solo dev, so don't hesitate.

Scraping Google instead? The Google Images Scraper pulls up to tens of thousands of full-size images per keyword.

• Thodor