Bing Search API Key in 2026: It's Retired — Here's What to Use

The Bing Search API v7 was retired on August 11, 2025. If you found this page Googling for “how to get a Bing Search API key,” the short answer is that you can’t — the API doesn’t exist anymore. The path forward depends on what you actually needed Bing data for.

This guide covers what happened to the API, Microsoft’s official replacement (and why it’s typically the wrong answer), the five real third-party alternatives, and what to do with the legacy v7 code that’s still in production at thousands of companies.

Timeline: how the retirement happened

The shutdown moved faster than most enterprise migrations. Per bitbasti’s coverage and PPC Land’s analysis:

Date	Event
February 2025	Microsoft quietly disabled new resource creation for Bing Search v7 in Azure
May 13, 2025	Microsoft formally announced retirement of Bing Search v7 and Custom Search APIs
August 11, 2025	Public endpoints stopped responding — return HTTP 410 Gone

Two days of formal notice between the May announcement and a sweeping public-API shutdown was unusually aggressive even by hyperscaler-deprecation standards. Per Windows Forum’s analysis, the abrupt timeline left many enterprise teams scrambling to migrate production workloads with little planning runway.

What Microsoft says you should use instead

The official replacement is Grounding with Bing Search, a feature inside the Azure AI Agent Service. Per Searchcans’s 2026 retirement guide, it returns Bing-grounded results to AI agents — not direct SERP JSON. The architectural differences from v7:

• Agent-shaped, not SERP-shaped. Grounding returns search results consumed by a Microsoft-managed AI agent that answers a query, rather than letting you pull the raw SERP and process it yourself.
• Requires the full Azure Agents framework. Resource provisioning, role assignments, agent configuration, region selection — none of which v7 required for a simple GET /search call.
• 40-483% more expensive depending on workload per PPC Land’s pricing analysis. Smaller-volume callers see the larger percentage increase.
• The underlying Azure Assistants API is itself in preview and scheduled for deprecation through the first half of 2026 per bitbasti’s coverage. The replacement runs on a platform that’s also being replaced.

For teams that just want Bing SERP JSON — the actual v7 use case — Grounding with Bing Search is structurally the wrong tool. It’s an AI-agent platform, not a search API. The realistic migration path for most teams is a third-party SERP API.

The 4 real Bing Search API alternatives in 2026

Vendor	Best for	Bing parsing	Other engines	Starting price
SerpApi	Broadest engine catalog	✅	80+ engines incl. Google, Baidu, Yandex, YouTube	$75/mo (5k searches)
DataForSEO	Lowest base SERP rate at scale	✅	Google, Bing, Yahoo, Yandex, Baidu, more	$0.60/1k (Standard Queue)
Brave Search API	Independent index alternative	n/a (own index)	Brave’s 30B+ page index	$5/1k (free tier available)
Firecrawl	AI-application curated search	partial	Curated authoritative-source index	Flat predictable pricing

The four split on whether you need actual Bing SERP data (SerpApi and DataForSEO do; Brave and Firecrawl substitute different indices). Below the table, we cover a fifth option — cloro — but as a different category answer: cloro doesn’t scrape Bing directly, it covers the AI search surfaces where a meaningful share of Bing-style use cases have migrated.

1. SerpApi — best for breadth across vertical engines

Per SerpApi’s documentation, the Bing endpoint returns parsed JSON for organic, ads, news, image, and video results. SerpApi’s bigger value is the catalog — 80+ search engines including Bing alongside Baidu, Yandex, Naver, eBay, YouTube, Amazon, Walmart — under one auth.

Pricing: Developer $75/month (5,000 searches, $15/1k effective). Production $150/month (15,000 searches). Higher tiers reduce the per-search rate.

2. DataForSEO — best low base rate at production scale

DataForSEO’s SERP API covers Bing alongside Google, Yahoo, Yandex, Baidu, Naver, and others. The September 2025 SERP-API price cut took the base rate down roughly 80%; the Standard Queue tier now runs $0.60/1k with ~5-minute queue latency. Live mode runs $2/1k. AI Overview enrichment ($2 → $4 on first page) doesn’t apply to Bing (no AIO equivalent on Bing).

Pricing: Pay-as-you-go from $0.60/1k (Standard Queue) to $2/1k (Live). No monthly minimum.

See our DataForSEO alternatives breakdown for the full DataForSEO trade-off analysis.

3. Brave Search API — best independent index

Brave runs its own 30B+ page search index, fully independent of Google and Microsoft. For teams that picked Bing specifically because they wanted a non-Google alternative, Brave is the closest spiritual successor — a different index with different ranking signals and different content surfacing.

Pricing: Free tier (2,000 searches/month). Paid plans from $5 per 1,000 searches.

Trade-off: It’s a different index, not Bing. Results won’t match what your existing Bing-tuned workflows expect. Best when “non-Google search index” was your goal, not “Bing specifically.”

4. Firecrawl — best for AI-application use cases

Firecrawl maintains a curated search index of authoritative sources (news, research, finance, government) with freshness monitoring. For RAG pipelines, AI agents, or research applications that need current full-content results rather than stale snippets, Firecrawl’s curated approach often beats a raw SERP. It combines search, scrape, crawl, and extraction in a single API with flat predictable pricing.

Trade-off: Not a Bing SERP replacement — it’s a different product category. If your Bing v7 use case was feeding an AI workflow rather than doing classic SEO rank tracking, Firecrawl may be the closer functional fit than any of the four above.

Adjacent option: cloro (for AI surfaces, not Bing directly)

cloro doesn’t scrape Bing.com. The covered surfaces are Google Search, Google AI Overview, Google AI Mode, ChatGPT, Perplexity, Gemini, Copilot, and Grok. Including it here is a category-aware honest answer to one specific Bing v7 use case: teams that were using Bing for AI-feeder workflows (RAG pipelines, agent data ingestion, AI brand-visibility tracking). Most of that use case has migrated from Bing-the-web-search-engine to the AI search surfaces themselves — ChatGPT, Perplexity, Copilot — which is what cloro covers.

There’s one place cloro touches the Bing ecosystem indirectly: Microsoft Copilot, which is built on Bing’s index. cloro’s Copilot endpoint returns Copilot’s responses with the Bing-grounded source URLs Copilot cites. That captures the slice of Bing-mediated traffic that now flows through Copilot’s AI conversational interface, but it is not the same product as scraping Bing.com directly for organic, ads, news, image, and video results.

Bottom line on cloro for Bing migration:

• ✗ Replacing Bing v7’s organic/web/news/image/video SERP scraping — use SerpApi or DataForSEO instead
• ✗ Replacing Bing v7 for SEO rank tracking on Bing.com — use SerpApi or DataForSEO instead
• ✓ Replacing the Bing-feeds-AI-workflow use case via Microsoft Copilot, ChatGPT, and Perplexity — cloro fits

What if you have legacy v7 code still running?

Pre-retirement Bing Search v7 calls now return HTTP 410 Gone. Production systems still hitting the old endpoints have been making failed calls since August 11, 2025. Three patterns we’ve seen:

1. Silent failure with logged errors. Most common. The integration logs the 410 but the surrounding workflow had no alerting on it; data quality degrades over weeks until someone notices a downstream report is empty.
2. Hard failure with bubbled exceptions. Less common but easier to diagnose. The integration crashes and someone fixes it immediately.
3. Cached results. Workflows that cached results aggressively may still appear to work for cached queries; new queries fail silently.

If you find your code in pattern 1, the cleanest migration is to a third-party SERP API that actually scrapes Bing. SerpApi’s Bing endpoint is the most v7-compatible in field naming; DataForSEO’s Bing endpoint is the cheapest at scale.

Legacy code samples (for reference only — these endpoints no longer respond)

Below are the canonical Bing Search v7 code patterns. They no longer work, but if you’re trying to diagnose what your existing integration was doing before migrating, this is the shape.

# Legacy Bing Search v7 — DO NOT USE in production after August 11, 2025
# Returns HTTP 410 Gone since retirement
import requests

API_KEY = "YOUR_KEY_FROM_BEFORE_RETIREMENT"
ENDPOINT = "https://api.bing.microsoft.com/v7.0/search"

headers = {"Ocp-Apim-Subscription-Key": API_KEY}
params = {"q": "best running shoes", "mkt": "en-US"}

response = requests.get(ENDPOINT, headers=headers, params=params)
# Will return 410 Gone since 2025-08-11

The migration target for that exact pattern, using SerpApi’s Bing endpoint (the closest field-naming match to v7):

# SerpApi Bing endpoint
import requests

SERPAPI_KEY = "YOUR_SERPAPI_KEY"
ENDPOINT = "https://serpapi.com/search.json"

params = {
    "engine": "bing",
    "q": "best running shoes",
    "cc": "US",
    "api_key": SERPAPI_KEY,
}

response = requests.get(ENDPOINT, params=params)
data = response.json()
# data["organic_results"], data["ads"], data["news_results"], etc.

The structural differences from v7:

• engine=bing query param. Specifies the search engine; the same auth and base URL handle Google, Bing, YouTube, and others.
• api_key in query string vs subscription key header. SerpApi uses a query-string key by default; can be moved to an Authorization header.
• Parsed result envelope. organic_results, ads, news_results arrays at the top level rather than v7’s nested webPages.value.
• cc parameter vs mkt. ISO country code rather than the locale-market combination v7 used.

Bing market context and why this matters

Even with the API retired, Bing remains a meaningful search surface, and the share is hidden in surprising places. Per Backlinko’s 2026 Bing usage stats and Searchlab’s 2026 search engine market share data:

• Bing holds 4.31% global market share across all devices and is the second-most-popular search engine globally with steady ~0.36% YoY growth
• Desktop share is 10.5% — roughly 17× the mobile share of 0.62% — thanks to Windows and Edge default-search integration
• 34% of Bing users don’t know they’re using Bing. They search via the Windows taskbar, Edge address bar, or Cortana, never visiting bing.com directly
• Microsoft Copilot uses Bing’s index for grounding and is integrated by default in Windows 11 and Edge. Every Copilot search query is technically a Bing search, and Copilot’s reach via OS defaults is growing faster than direct Bing.com visits

The 34% “don’t know they’re using Bing” number is the most consequential one for migration planning. It means brand-monitoring or SEO-tracking workflows that drop Bing entirely are missing a chunk of search traffic that users themselves don’t classify as a separate channel. From the user’s perspective they’re just “searching from Windows.” From the data-stack perspective, it’s Bing.

For teams doing brand monitoring, competitive intelligence, or SEO tracking, dropping Bing from the data stack means dropping a ~4-10% slice of the global search surface (higher on desktop, dominant inside Windows ecosystems) plus an outsized share of Copilot-mediated AI traffic. The third-party SERP APIs above are the practical migration path; Grounding with Bing Search in Azure AI Agents is the official path that most teams find structurally wrong.

Decision tree: pick the right replacement

Use this:

• Need Bing SERP JSON with a similar shape to legacy v7? SerpApi (field-naming closest) or DataForSEO (lowest base rate at scale). Both scrape Bing.com directly.
• Was Bing your “non-Google index” choice and you don’t actually need Bing-the-engine? Brave Search API — independent index, similar use case.
• Building an AI agent or RAG pipeline rather than doing classic rank tracking? Firecrawl — curated authoritative-source search is often the better functional fit.
• Was your v7 use case feeding an AI workflow (brand-visibility tracking, RAG, agent search)? Most of that traffic has migrated from Bing-the-web-search-engine to the AI surfaces themselves. cloro covers ChatGPT, Perplexity, Gemini, Microsoft Copilot (Bing-grounded), Google AI Overview, and AI Mode — not Bing directly, but the AI surfaces that absorbed the feeder use cases.
• Have to stay on Microsoft’s official path for compliance/contractual reasons? Grounding with Bing Search in Azure AI Agents — accept the 40-483% cost increase and the Azure Agents framework overhead.

The bottom line

The Bing Search API v7 is gone. Microsoft’s replacement is structurally an AI-agent product, not a search API, and runs significantly more expensive. For the SEO, monitoring, and intelligence-gathering use cases v7 served on Bing-the-search-engine directly, the practical 2026 answer is SerpApi or DataForSEO. For the AI-workflow-feeder use cases that v7 served (RAG pipelines, agent data ingestion, brand-visibility tracking), the right answer is usually the AI-search-surface APIs themselves — cloro for ChatGPT, Perplexity, Gemini, Copilot, AI Overview, and AI Mode coverage; Firecrawl for curated full-content search.

If your Bing v7 workload was the AI-feeder pattern, start with cloro — 500 free credits is enough to test how the AI surfaces actually handle your priority queries before committing. If you need Bing.com data specifically, SerpApi or DataForSEO are the direct replacements.

For the deeper SERP-API category comparison see Best SERP APIs in 2026. For pricing specifically see Cheapest SERP API 2026. For Google-specific scraping see Best Google Scraper 2026.