Bing Images API: How to Extract Image Results as JSON

Bing Images is a useful source for visual search data.

For developers, SEO teams, AI apps, and data teams, image search results can reveal how topics, products, brands, people, places, and visual concepts appear across the web. Instead of manually opening Bing Images and copying links, a Bing Images API lets you collect image results as structured JSON.

This is useful when you need:

• Image titles
• Thumbnail URLs
• Original image URLs
• Source page URLs
• Source domains
• Image dimensions
• Ranking positions
• Related visual results
• Localized image search data

TalorData provides a Bing Images API for collecting real-time image search results, including titles, source URLs, thumbnails, original image links, dimensions, ranking positions, and related visual results. The API supports structured JSON or HTML output and geo-location targeting.

A basic workflow looks like this:

Search query
↓
Bing Images API
↓
Structured JSON response
↓
Image result extraction
↓
Database, dashboard, AI workflow, or visual SEO report

One important note: Microsoft retired its Bing Search APIs on August 11, 2025, and decommissioned existing instances. Microsoft now points customers toward Grounding with Bing Search inside Azure AI Agents. For teams that still need structured Bing image search results as JSON, a SERP API workflow is often the practical path.

What Is a Bing Images API?

A Bing Images API lets developers send a search query and receive image search results in a machine-readable format.

Instead of parsing a visual search page manually, you can request structured data and extract fields such as image title, thumbnail, source URL, and ranking position.

Example use cases include:

The goal is not just to collect image links.

The goal is to turn image search results into structured data that can be stored, filtered, compared, and reused.

What Image Data Can You Extract?

A useful Bing Images JSON response should include fields like:

TalorData’s Bing Images API is designed to collect image results, thumbnails, source pages, original image URLs, dimensions, visual tags, and localized image results.

A normalized image result can look like this:

{
  "query": "modern office projector",
  "position": 1,
  "title": "Modern projector in office meeting room",
  "thumbnail_url": "https://example.com/thumb.jpg",
  "image_url": "https://example.com/full-image.jpg",
  "source_url": "https://www.example.com/projector-guide",
  "source_domain": "example.com",
  "width": 1200,
  "height": 800,
  "collected_at": "2026-07-17T09:00:00Z"
}

This structure is easier to use than raw HTML because every field has a clear purpose.

Step 1: Define the Search Query

Start with a focused query.

Bad query:

shoes

Better query:

white running shoes product photography

For image search, query wording affects visual intent. A broad query may return mixed results, while a specific query helps you collect cleaner image data.

Useful query patterns include:

For production workflows, store the query with every result. Without the query, the image record loses context.

Step 2: Add Location and Language Context

Image results can vary by market.

A Bing Images API workflow should support localization parameters such as country, language, or region. TalorData’s Bing Images parameter documentation covers query configuration, localization, geographic targeting, and advanced output options.

Useful parameters include:

A request pattern may look like this:

{
  "engine": "bing_images",
  "q": "modern office projector",
  "country": "us",
  "language": "en",
  "device": "desktop",
  "output": "json"
}

Use the exact endpoint and parameter names from your TalorData dashboard or API documentation. The important idea is to make image search reproducible.

Step 3: Extract Image Results from JSON

Once the API returns JSON, extract only the fields your workflow needs.

Example Python pattern:

from urllib.parse import urlparse
from typing import Any

def get_domain(url: str) -> str:
    if not url:
        return ""
    return urlparse(url).netloc.replace("www.", "")

def normalize_image_results(
    response: dict[str, Any],
    query: str,
) -> list[dict[str, Any]]:
    raw_results = (
        response.get("image_results")
        or response.get("images")
        or response.get("results")
        or []
    )

    normalized = []

    for index, item in enumerate(raw_results, start=1):
        source_url = item.get("source_url") or item.get("source") or item.get("link") or ""
        image_url = item.get("image_url") or item.get("original") or item.get("content_url") or ""
        thumbnail_url = item.get("thumbnail_url") or item.get("thumbnail") or ""

        normalized.append(
            {
                "query": query,
                "position": item.get("position") or index,
                "title": item.get("title") or item.get("name"),
                "thumbnail_url": thumbnail_url,
                "image_url": image_url,
                "source_url": source_url,
                "source_domain": item.get("source_domain") or get_domain(source_url),
                "width": item.get("width"),
                "height": item.get("height"),
            }
        )

    return normalized

The field names in your real response may differ. Keep your parser small, visible, and easy to update.

Step 4: Store Image Results as Snapshots

For monitoring workflows, do not overwrite old results.

Store snapshots.

A simple table can include:

Snapshots allow you to answer questions like:

• Which domains appear most often?
• Which images disappeared from the results?
• Which brand visuals became more visible?
• Which product images rank higher over time?
• Which thumbnails changed?

This matters for visual SEO, brand monitoring, and e-commerce image analysis.

Step 5: Filter and Deduplicate Results

Image search often returns duplicates or near-duplicates.

Before sending results to a dashboard or AI workflow, filter them.

Useful rules:

For AI workflows, filtering is especially important. Agents do not need 100 similar thumbnails. They need a clean set of useful image records.

Step 6: Use Bing Images JSON in AI Workflows

Structured image results can support AI workflows without requiring the model to browse visual pages manually.

Example workflow:

User asks for visual research
↓
Bing Images API returns JSON
↓
Workflow filters image results
↓
AI summarizes visual patterns
↓
Selected source URLs are stored or reviewed

This can support:

The AI should not treat every image URL as verified truth. Use image search data as discovery context, then review or fetch source pages when needed.

How TalorData Helps

TalorData Bing Images API helps developers collect real-time image search data as structured JSON, including image titles, thumbnails, source URLs, original image links, dimensions, ranking positions, and related visual results. It supports geo-location targeting, successful-request-based billing, and JSON or HTML output.

With TalorData, the workflow becomes:

Search query
↓
TalorData Bing Images API
↓
Structured JSON
↓
Filtered image records
↓
Database, dashboard, AI agent, or report

This is useful for teams building visual SEO tools, brand monitoring systems, e-commerce research workflows, AI agents, and image-based market research products.

Final Thoughts

A Bing Images API is useful when image search results need to become structured data.

The core workflow is simple:

Define the query.
Set location and language.
Request Bing Images results.
Extract titles, thumbnails, source URLs, and image URLs.
Normalize the JSON.
Store snapshots.
Filter duplicates.
Use the data in dashboards, reports, or AI workflows.

For developers, the value is not just getting images.

The value is getting image search data in a format your systems can actually use.

FAQ

What is a Bing Images API?

A Bing Images API lets developers collect Bing image search results as structured data, including titles, thumbnails, image URLs, source pages, dimensions, and ranking positions.

Is Microsoft Bing Image Search API still available?

Microsoft retired Bing Search APIs on August 11, 2025, and decommissioned existing instances. Developers who need structured image search results can use a SERP API workflow.

What fields should I extract from Bing image results?

At minimum, extract title, thumbnail URL, original image URL, source URL, source domain, ranking position, dimensions, query, location, language, and collection time.

Can Bing Images data be used in AI apps?

Yes. Structured image results can support AI visual research, brand monitoring, product research, content planning, and RAG source discovery.