Bing Search API Key in 2026: Retired — Use These Instead

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.

Can you still get a Bing Search API key?

No — and the reason matters for your migration. There is no supported way to generate a new Bing Search API key in 2026, and any key issued before retirement no longer returns data.

How you used to get a Bing Search API key

Before the shutdown, getting a Bing Search API key meant provisioning a “Bing Search v7” resource in the Azure Portal. You created the resource, opened its Keys and Endpoint blade, and copied one of the two subscription keys into the Ocp-Apim-Subscription-Key request header. The F1 free tier included 1,000 transactions per month. Paid S-tiers billed per 1,000 transactions above that.

Microsoft disabled new Bing Search resource creation in February 2025. From that point you could no longer generate a fresh Bing Search API key at all. Any Bing Search API key provisioned earlier kept working only until August 11, 2025, when the endpoints began returning HTTP 410 Gone. So if your goal today is simply to get a Bing Search API key, there is no supported path — the rest of this guide covers what to use instead.

The phrase still gets searched constantly. Developers inherit an old codebase with a hard-coded Bing Search API key, hit a wall, and search for how to renew or replace it.

The honest answer is that the credential itself is dead. No dashboard will issue a new one, and no support ticket will reactivate an old one. Your effort is better spent choosing a replacement API than chasing a Bing Search API key that Microsoft has permanently decommissioned.

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 OpenAI Assistants API is itself deprecated, with a hard retirement date of August 26, 2026 and developers pointed to Microsoft’s Foundry Agents service (built on the Responses API). The official replacement runs on a platform that is itself being replaced on the same timeline.

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 ($5 monthly credits)
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. For buyers evaluating SerpApi against managed competitors at production volume, see our SerpApi alternatives breakdown.

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: $5 per 1,000 searches on Brave’s independent index. Per Brave’s API pricing, there’s no standing free tier — every plan instead includes $5 in free credits each month, applied automatically. Brave also exposes dedicated news, image, and video endpoints on the same index.

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 (Google SERP + 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. It’s a category-aware honest answer to two Bing v7 use cases that were never really about Bing:

• You just needed general web search. A large share of teams reached for the Bing Search API because Google never shipped an official one — not because they needed Bing’s index specifically. If that was you, cloro’s live Google SERP is the direct swap: organic results, ads, AI Overview, and People Also Ask in one JSON call, against a bigger index than Bing’s. It’s the strongest replacement on this page for the “I just want web-search results” case — as long as you don’t need Bing’s own ranking.
• You were feeding an AI workflow (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 exactly 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 when what you actually needed was general web search and Bing was just the API you could get — cloro’s live Google SERP is the direct swap, and a bigger index
• ✓ Replacing the Bing-feeds-AI-workflow use case via Microsoft Copilot, ChatGPT, and Perplexity — cloro fits
• ✗ Replacing Bing when you need Bing’s own results — organic/ads/news/image/video from Bing’s index, or Bing.com rank tracking — use SerpApi or DataForSEO instead

Bing Search API pricing: what v7 cost vs. what alternatives cost now

Two of the most-searched follow-ups after the shutdown are “Bing Search API pricing” and “Bing Search API documentation” — usually from someone who inherited a v7 integration and now has to budget the replacement. Here’s the honest before-and-after.

What Bing Search v7 cost. The F1 tier gave you 1,000 free transactions per month. Above that, paid S-tiers billed per 1,000 transactions, and the rate climbed sharply over the API’s life. After Microsoft’s February 2023 price increase, the standard tiers landed at $25/1,000 (S1), $15/1,000 (S2), and $6/1,000 (S3) per PPC Land’s pricing analysis. Those are the numbers your old Azure invoices reflect.

What the official replacement costs. Grounding with Bing Search in Azure AI Agents runs $35 per 1,000 transactions — a 40% increase over the old S1 rate and a 483% jump over the S3 rate heavier-volume teams were on, per the same analysis. That’s the “40–483% more expensive” headline in one line: same query, more money, and an agent-shaped response instead of raw SERP JSON.

What third-party SERP APIs cost. For actual Bing SERP data, the drop-in replacements undercut both Microsoft options:

Option	Cost per 1,000	Response shape
Bing v7 (retired)	$6–$25	Raw SERP JSON
Grounding with Bing (official)	$35	Agent answer, not SERP
DataForSEO	$0.60 (Standard Queue)	Raw Bing SERP JSON
SerpApi	~$15 ($75/mo for 5k)	Raw Bing SERP JSON
Brave Search API	$5	Independent index (not Bing)

At DataForSEO’s $0.60/1,000, the same workload that cost $6–25/1,000 on Bing v7 — and would cost $35/1,000 on Grounding — runs roughly 10–40× cheaper, with a SERP-shaped response that matches what v7 returned.

On documentation: the old learn.microsoft.com Bing Search v7 reference is being retired with the API; Microsoft’s retirement announcement now points to the Grounding with Bing Search docs instead. If you’re maintaining legacy code, the field-mapping notes later in this guide are more useful than the archived v7 reference, since the endpoints themselves return HTTP 410 Gone.

Bing Image Search API: what replaced it

The Bing Image Search API didn’t get a separate reprieve — it was retired in the same August 11, 2025 wave as Bing Web Search. Microsoft’s retirement announcement decommissioned the whole family at once: Web, Image, Video, News, Entity, Autosuggest, Spell Check, Visual Search, and Custom Search. If your app called /v7.0/images/search, it returns HTTP 410 Gone today.

Two practical replacements:

• SerpApi Bing Images API — the closest like-for-like. SerpApi kept a dedicated Bing Images endpoint that returns parsed image results (thumbnails, source URLs, dimensions) under the same auth as its other engines, and it’s the only one of these with a dedicated Bing reverse-image endpoint for visual-search workflows. The $75/month (5,000 searches) plan covers it.
• Pivot to Google Images — if the image results themselves mattered more than the fact they came from Bing, a Google Images SERP endpoint (SerpApi or DataForSEO) taps a larger index. cloro covers Google Search and the AI answer surfaces rather than a standalone image-search endpoint, so for image-specific SERP data SerpApi or DataForSEO are the fit.

Which to pick depends on whether “Bing” was load-bearing. For reverse-image or visual-search workflows specifically, SerpApi is the one to reach for.

Bing News Search API: what replaced it

Same story as images: the Bing News Search API was decommissioned in the August 11, 2025 retirement, not spun out as a survivor. Calls to /v7.0/news/search now return 410.

The replacements, ranked by fit:

• SerpApi Bing News API — a dedicated Bing News endpoint returning parsed articles (headline, source, timestamp, snippet). Closest to what v7’s news vertical returned.
• DataForSEO — Bing SERP endpoints surface news blocks inline, and it’s the cheapest at scale ($0.60/1,000 Standard Queue).
• Google News as a substitute index — if you wanted news coverage rather than Bing’s news ranking specifically, a Google News feed is the broader index. See our Google News API guide for what exists there and what to use instead.

For most teams the news vertical was about monitoring coverage, not about Bing’s specific ranking — which makes a Google-News-backed feed the more complete replacement than a like-for-like Bing News scrape.

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.
• Did you just need general web-search results, and Bing was only the API you could get? cloro — live Google SERP (organic, ads, AI Overview, People Also Ask) in one JSON call, against a bigger index than Bing’s. Not Bing’s own results, but for general web search that’s usually an upgrade.
• 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 that genuinely need Bing’s own results, the practical 2026 answer is SerpApi or DataForSEO. But if Bing was only incidental — you needed general web search and Bing was just the API you could get — cloro’s live Google SERP is usually the better swap, against a bigger index. And for the AI-workflow-feeder use cases 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 general web search or the AI-feeder pattern, start with cloro — 500 free monthly credits is enough to test how Google SERP and the AI surfaces handle your priority queries before committing. If you need Bing’s own results 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.