Bright Data pricing is a catalog, not one sticker. On 29 August 2026 the public SERP API and Web Unlocker pages both start at 5,000 free requests, then $1.50 per 1,000 on pay as you go, then a $499 Scale plan. The Scale included-request counts are not the same: SERP lists 380,000, Unlocker lists 383,000. MCP Server Pricing reprints the $1.50 / $1.30 / $1.10 / $1.00 per 1,000 scrape rates plus $8 / $7 / $6 / $5 per GB for browser navigation, with Starter at $499, Professional at $999, and Business at $1,999 billed monthly. Bright Data wins if the job is scraping live pages or SERP HTML. Trends MCP wins if the job is ranked trend boards and keyword growth over MCP plus REST. It is not a proxy network.

What the public catalog actually sells

The pricing hub (schema dateModified 23 August 2026) lists eight products: Proxy Networks, Web Scraper API, Scraper Studio, Scraping Browser, SERP API, Web Unlocker, Datasets, and Bright Insights. Each row has its own "See Pricing" page. A buyer who Googles "Bright Data pricing" and expects a single SaaS grid will mis-budget.

Signup copy on the SERP, Unlocker, MCP, and Google Trends product pages still offers to match the first deposit dollar for dollar, up to $500. Charges are billed in USD; local-currency figures on those pages are estimates. Several product pages also print "Trusted by 50,000+ customers worldwide." Those claims are vendor marketing, not a third-party audit.

This post quotes only the SERP, Unlocker, and MCP pages plus the Google Trends SERP docs. Proxy GB rates, Scraping Browser minutes, Datasets subscriptions, and Insights seats were not copied from those other grids, so they are omitted. Check current pricing on each product page before a purchase order.

SERP and Unlocker Scale rows disagree by 3,000 requests

SERP API Pricing (schema dateModified 25 August 2026) prints four self-serve rows:

PlanWhat the page printsExtra
Free Tier5,000 requests per month, no cardExpert support
Pay as you go$1.50 per 1,000 requestsPay only for success, spend limits, unlimited concurrency
Scale$499 per month, 380,000 requests included$1.30 per 1,000 additional, cancel anytime
EnterpriseCustomVolume discounts, account manager, premium SLA, priority support, SSO

Web Unlocker API Pricing (schema dateModified 25 August 2026) looks identical until Scale: that row includes 383,000 requests, not 380,000. Free, pay as you go, and the $1.30 overage rate match SERP. Do not copy one table onto the other.

SERP FAQ copy on that page: each API execution counts as a request; in asynchronous mode only the send is counted. Failed deliveries are not billed (repeated on the Google Trends SERP product page). Concurrent request cap: none. Geo coverage on the same FAQ: 195 countries plus city-level targeting. Response-time claims on the Trends product page conflict with each other: the hero says "under 1 second," a later block says "under 5 seconds," and the SERP FAQ says "under 5 sec." Treat latency as unverified marketing until a contract SLA is in writing.

MCP is Unlocker credits with a 69-tool catalog

Docs fetched 29 August 2026 state the MCP server runs on the Web Unlocker API. The 5,000 free monthly requests are the same account-level pool, shared across team users, not a second grant. When that pool is exhausted, requests start failing until an Unlocker zone is created and passed as &unlocker=ZONE_NAME on the remote URL or WEB_UNLOCKER_ZONE on local npx.

The MCP pricing page (schema dateModified 8 January 2026, older than the SERP and Unlocker grids) adds browser-navigation GB rates on top of the per-1,000 scrape rates:

PlanSearch / scrape / extractBrowser navigationMonthly
Free5,000 monthly requestsMarked as available in paid tiers on the card$0
Pay as you go$1.50 per 1,000$8 per GBNo commitment
Starter$1.30 per 1,000$7 per GB$499 billed monthly
Professional$1.10 per 1,000$6 per GB$999 billed monthly
Business$1.00 per 1,000$5 per GB$1,999 billed monthly

The same page asks "Need more than 2M requests?" and points to sales, with 99.99% network uptime listed as a custom-tier claim. Starter / Professional / Business do not print an included-request bucket the way SERP Scale prints 380,000. Do not invent one.

Tool counts also split across Bright Data's own properties. The GitHub README fetched this run says the server exposes 69 tools. Remote quickstart and FAQ pages say 60+ tools. Pricing-page FAQ still tells operators to create a Browser API zone and set BROWSER_AUTH. Current docs FAQ says to enable browser tools with &pro=1 or &groups=browser on the remote URL, or PRO_MODE=true / GROUPS=browser locally. Cite both paths; the older FAQ is still live.

Remote endpoints documented 29 August 2026:

https://mcp.brightdata.com/sse?token=YOUR_API_TOKEN
https://mcp.brightdata.com/mcp?token=YOUR_API_TOKEN

Cursor config from the same README uses url only. No transport field. The token sits in the query string, not an Authorization header:

{
  "mcpServers": {
    "brightdata": {
      "url": "https://mcp.brightdata.com/mcp?token=YOUR_API_TOKEN"
    }
  }
}

Local stdio still uses npx @brightdata/mcp with API_TOKEN in env. Optional URL flags on the hosted server include groups= and tools= to shrink the tool list. Batch tools search_engine_batch and scrape_batch cap at 10 items. README timeout guidance for slow web_data_* calls is 180 seconds in the agent.

Bright Data's Google Trends product is a SERP API target, not a separate SKU. It inherits the SERP grid above. Docs at SERP API Google query parameters (fetched 29 August 2026) require brd_json=1 (parsed JSON only) and recommend brd_trends=timeseries,geo_map. The same section then says Bright Data supports only the "Trending now" tab; Explore and Home "are not supported as they require login access for full functionality."

The examples under that warning still hit trends.google.com/trends/explore with timeseries and geo_map widgets, date presets (now 1-H through today 5-y plus custom ranges), cat, gprop (images, news, froogle, youtube), and hl. The product page curl samples also use a different Super Proxy port (33335) than the docs samples (44445). Anyone buying this for Google Trends Explore history should ticket Bright Data before budgeting; the public docs disagree with themselves.

Trends MCP does not scrape trends.google.com. It serves a get_top_trends feed named Google Trends (and Google Trends by Category with official names such as Games) plus keyword get_growth / get_time_series on google search. Default limit is 25, max 200, billed per feed and per page. That is a different product. Details sit on the Google Trends MCP for Cursor page.

Trends MCP is cheaper than Bright Data Scale only when the workload is trend boards and keyword series, not HTML scraping. HTML /pricing fetched 29 August 2026 (~16:15 UTC) printed 9,486,262 API calls served and these self-serve rows:

PlanMonthly requestsMonthly cardYearly as billed
Free100$0$0
Starter1,000$19$16/mo, $192/year
Pro5,000$49$39/mo, $468/year
Business25,000$199$159/mo, $1,910/year

Yearly toggle copy: "2 months free" (20% off). Paid plans collect a card at checkout, start a 7-day trial, and charge on day 8. Failed calls are free. Hitting the quota returns 429; there is no overage meter. Free is capped to top 10 trends, 12 months of history, and 24-hour delayed data. Paid rows list all trends, full history, and real-time. Keyword alerts: 50 on Pro, 250 on Business.

Source counts must not be flattened. HTML markets 35+ data sources. docs.md the same day lists 15 keyword sources (no python) and 24 live feed names, including TikTok Trending Hashtags by Category and Amazon Best Sellers by Category. data-sources.md lists those 15 keyword sources and only 20 live feeds, omitting all four by Category boards. llms.txt says plus 21 live feeds. The MCP get_top_trends schema this run lists 24 feeds. get_growth still accepts python; the docs keyword table does not. Cite the split.

Cursor for Trends MCP uses url, transport: http, and a Bearer header against api.trendsmcp.ai. Claude.ai Connectors use OAuth on the www /mcp host instead. Exact JSON:

"trends-mcp": {
  "url": "https://api.trendsmcp.ai/mcp",
  "transport": "http",
  "headers": { "Authorization": "Bearer YOUR_API_KEY" }
}

Setup notes live on the MCP server for Cursor page. A wider vendor grid is on trends API pricing comparison. get_growth counts as one request per source plus keyword even when several period presets ride in one call.

JobBright DataTrends MCP
Scrape an arbitrary URL as Markdown/HTMLMCP scrape_as_markdown / UnlockerNot the product
Live Google / Bing / Yandex SERP JSONSERP API, 195 countries, city geoNot the product
Ranked Google Trends board, default 25 rowsSERP Trends target; docs conflict on Explore vs Trending nowget_top_trends type Google Trends
Multi-source keyword growth (TikTok, Amazon, npm, Steam)Scrape each site yourselfget_growth with comma-separated source
Cursor installToken in url query, no transporturl + transport: http + Bearer header

Bright Data remains the pick for unblocking, browser automation, and SERP HTML at volume. Trends MCP is the pick for named trend feeds and indexed keyword series without standing up zones.

Frequently asked questions

Is there a single Bright Data price?

No. The hub lists eight products. SERP and Web Unlocker share a $1.50 per 1,000 pay-as-you-go rate and a $499 Scale plan, but Scale includes 380,000 SERP requests versus 383,000 Unlocker requests. MCP reprints the scrape rates and adds GB billing for browser navigation ($8 down to $5 per GB). Proxy, Datasets, and Insights are separate checkouts.

Does the MCP free tier stack on top of Unlocker?

No. Docs say it is the same 5,000-credit Unlocker pool, shared across the account. After that, remote calls need an Unlocker zone name on the URL. The GitHub README still advertises 5,000 requests per month with no card.

Can Bright Data replace Google Trends Explore?

Not on the public docs as written. The Google query-parameter page says only the Trending now tab is supported, then prints Explore URLs with timeseries and geo_map. Ticket the vendor before treating Explore history as a billed SKU. Trends MCP does not scrape Explore either; it returns its own Google Trends board and google search series.

When is Trends MCP the cheaper bill?

When the agent needs ranked boards and keyword growth, not page HTML. Starter is $19 for 1,000 requests versus Bright Data Scale at $499. That comparison is invalid for scraping Amazon PDP JSON or solving CAPTCHAs. Bright Data's 5,000 free Unlocker/MCP requests beat Trends MCP's 100 free requests if the work is web fetch.

Do failed Bright Data SERP calls cost money?

The SERP and Google Trends product FAQs say no: only successful deliveries bill. Unlocker marketing repeats "pay only for success." Confirm the rule on the invoice, not the landing page, for a production workload.

Accessed 29 August 2026: Bright Data pricing hub, SERP API pricing, Web Unlocker pricing, MCP Server pricing, Google Trends SERP product, SERP Google query parameters, MCP overview, MCP FAQs, brightdata-mcp README. Trends MCP HTML /pricing, pricing.md, docs.md, data-sources.md, and llms.txt fetched the same day.