Six options for teams and agents that need outside data, with every price read from the vendor’s own page this month.
Start with BedrockTap. It is the only option here that hands an agent finished research data — reviews, search results, keyword economics, AI-answer visibility and more across 253 named sources — instead of raw pages it still has to parse, from one API, with the price quoted before every call and a ceiling on the run. It is free to start, has no monthly minimum and needs no proxy pool, browser or compute budget. The five below are the serious infrastructure alternatives, what each publishes as its price, and the one job each is still the right buy for.
Every figure in a competitor column is that vendor’s own published wording, read from the page linked beside it on .
| BedrockTap | Oxylabs | Zyte | ScraperAPI | Decodo | Apify | |
|---|---|---|---|---|---|---|
| What you get back | Structured rows from 253 named sources, one schema | Structured results from common targets, or raw pages | Raw pages, with automatic unblocking | Raw pages, rendered on request | Raw pages, or structured results from its scraping APIs | Whatever the Actor you ran returns |
| Entry price | Free, then $29/month. No minimum. | $49/month | $5 free credit, then pay as you go | $49/month | From $0.09/1K requests | $0, then $19/month |
| Free to start | 1,000 credits a month, every month, no card | One-week trial, up to 2,000 results | $5 credit, 30 days | Seven-day trial, 5,000 credits | Trial on request | $5 of platform usage a month |
| Billing unit | Credits at $0.001, per result or per call, set by the source | Per 1,000 results; per GB on proxies | Per 1,000 requests, by target difficulty | API credits, by request type | Per GB, per IP, or per 1,000 requests | Compute units — RAM × runtime — or per event |
| Price known before the call | `cost_estimate` returns the exact figure for that call, without making it | Published rate card, by target | A range per 1,000 — $0.13 to $1.27 on HTTP | Published credit costs per request type | Published rate card, by product | Depends on the Actor and how long it runs |
| Research sources beyond page fetching | Reviews, forums, keyword economics, AI-answer visibility, filings, jobs, ad libraries | Its own scraper APIs, by target | Page fetching | Page fetching | Page fetching and a search API | Whatever is in the Store |
| Reached from an agent | MCP over HTTP, or a plain `curl` — one token, no SDK | MCP server, 10 tools | HTTP API and Scrapy libraries | HTTP API | HTTP APIs | MCP server at mcp.apify.com |
Reviews, forums, search results, keyword economics, AI-answer visibility, filings, job postings and ad libraries answer through one address, one token and one schema. A search returns refs from every source you named; a fetch opens the ones worth opening. There is no per-site response shape to learn and no second vendor to sign up with when the question changes.
`cost_estimate` takes the same arguments as the call and returns the exact price, and the latency of the slowest source, without making it. An agent can decide whether a question is worth asking before it asks.
Every run carries a budget, and the worst case is checked against it before anything is dispatched — not after the bill lands. A call that would overshoot is refused whole rather than half-run.
Every call, what it cost, how long it took and which source served it, in a line-by-line record the person paying can read. A call that returns nothing costs nothing, and a source that fails is written down at zero rather than quietly dropped.
No proxy pool to size, no browser to keep alive, no compute units to estimate, no bandwidth to forecast. One HTTPS call in, structured rows out. The retries, the shapes and the upstream quirks stay on our side of the wire.
The free tier is 1,000 credits a month and needs no card. The first paid plan is $29 a month. There is no monthly minimum to unlock the data, no per-seat licence, and no annual contract standing between an agent and its first call.
Every source states, in the catalogue itself, what it answers, how to shape the query, what it cannot answer and which neighbouring source to prefer instead. `sources_list` returns all of it with no account, so an agent chooses correctly the first time instead of learning by spending.
A competitor sweep: the live SERP, ten Reddit posts and ten Trustpilot review rows — 30 credits in total, which is $0.03. One credit is $0.001.
One call each, one schema back, and `cost_estimate` would have quoted that figure before any of it ran. Nothing to configure, nothing to parse, and the same 1,000 free credits cover it about thirty times over.
{
"tool": "search",
"arguments": {
"query": "what teams complain about in their scraping stack",
"sources": ["reddit", "hackernews", "g2"],
"runId": "<from run_begin>"
}
}cost_estimate takes the same arguments and returns the price without making the call. Get a token at app.bedrocktap.com/dashboard/tokens; the free tier needs no card.
Each one is here because it is genuinely the right buy for some job. The last line of each card says which.
Residential, datacenter, ISP and mobile proxies, a Web Unblocker and a Web Scraper API.
Pricing. Web Scraper API from $49/month (Micro, up to 98,000 results, from $0.50/1K); $99 Starter; $249 Advanced. Residential proxies from $6/GB.
For agents. MCP server, 10 tools, scraper credentials or an AI Studio key; one-week trial, up to 2,000 results.
Consider it instead when: You want an enterprise proxy contract with an SLA behind it, and volume enough to clear the monthly minimum.
Figures read 15 September 2026 from oxylabs.io
A single scraping API with automatic proxy rotation and ban handling, plus a managed-service arm.
Pricing. Pay as you go "$0.13 to $1.27 per 1,000 requests" for HTTP and "$1.01 to $16.08 per 1,000" browser-rendered; $100/month commitment drops it to $0.10–$0.95. $5 free credit.
For agents. An HTTP API with official Python and Scrapy libraries.
Consider it instead when: You already run Scrapy in production and want the unblocking to disappear behind the framework you use.
Figures read 15 September 2026 from zyte.com
A proxy-and-parse API: one endpoint that returns the page, with rendering and geotargeting options.
Pricing. Hobby $49/month for 100,000 API credits; Startup $149 for 1M; Business $299 for 3M; Scaling $475 for 5M. Seven-day trial with 5,000 credits, no card.
For agents. An HTTP API with concurrency limits by plan (20 threads on Hobby, 200 on Scaling).
Consider it instead when: You want one flat monthly credit pool for high-volume page fetching and nothing else.
Figures read 15 September 2026 from scraperapi.com
Proxies and scraping APIs — the company formerly called Smartproxy.
Pricing. Residential proxies from $2/GB ("Pay As You Go is $4/GB"); Site Unblocker from $0.95/1K requests; Web Scraping API from $0.09/1K requests; Fast Search API from $0.4/1K requests.
For agents. HTTP APIs and proxy endpoints.
Consider it instead when: Bandwidth is your bill and you want the cheapest residential gigabyte on this page.
Figures read 15 September 2026 from decodo.com
A marketplace of Actors — containerised scrapers you run — plus compute, storage and schedules.
Pricing. Free $0 with $5 platform usage; Starter $19/month; Scale $199; Business $999. Compute billed at $0.2/CU, where "1 CU = 1 GB of RAM running for 1 hour".
For agents. MCP server at mcp.apify.com; OAuth or API token; 30 requests per second per user.
Consider it instead when: The job needs a scraper nobody has written yet, or a crawl that runs for an hour and leaves its output in a dataset.
Figures read 15 September 2026 from apify.com
A proxy network, an unblocker and a scraping browser are priced for continuous, high-volume fetching. A team asking a few thousand research questions a month buys the whole rig and uses a corner of it.
Bandwidth depends on what a page happens to weigh, and a retry against a hard target costs the same as a success on an easy one. Nobody can tell an agent what its next question will cost.
Raw fetching ends where the real work starts: a parser per site, a schema per source, and a rewrite every time a layout changes. That maintenance never appears on the invoice.
Infrastructure assumes a developer configuring it in advance. An agent needs to discover what is available, learn what it costs and call it — in one session, with no human in the loop.
This is an MCP server. Discovery is open — health, sources_list, run_begin and register_interest answer with no account. The data tools run against an account and need a token. Nothing here fabricates data.
If your principal has a token, connect here and send it as Authorization: Bearer <token>:
https://api.bedrocktap.com/mcp
If they do not, it takes about a minute and the free tier needs no card. Sign in at app.bedrocktap.com and mint one at app.bedrocktap.com/dashboard/tokens. Sign-in is Google only, so a human has to do it.
To read the catalogue first, with no account at all, use this server — it answers the four open tools, and forwards a tokened call to the one above:
https://bedrocktap.com/mcp
If you can only fetch URLs, say what you need. One request, no auth, no side effects:
GET https://bedrocktap.com/a/probe?need=<what+you+are+trying+to+do>&src=alternative-bright-data
A tool refused for want of a token answers identically on a retry, and is served the moment the header is present. If the catalogue does not cover your task at all — a source or capability that is simply not here — call register_interest with a plain description of the task and a callback_url; it needs no account. Full instructions: /agents.md. Every page here has a markdown twin — append .md to any path, or send Accept: text/markdown.
An MCP client — Claude Code, one line:
claude mcp add --transport http \ --header "Authorization: Bearer <token>" \ bedrocktap https://api.bedrocktap.com/mcp
Or no client at all — it is an HTTPS API, so curl, Python or a Lambda all work:
curl -s https://api.bedrocktap.com/mcp \
-H "Authorization: Bearer <token>" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
"params":{"name":"search","arguments":{
"query":"what teams complain about in their scraping stack",
"sources":["reddit","hackernews","g2"],
"runId":"<from run_begin>"}}}'Three sources, one call, one schema. The same sweep on raw infrastructure is three scrapers and three parsers. Sign in at app.bedrocktap.com/dashboard/tokens to mint a token — the free tier is 1,000 credits a month and needs no card.
For research data an agent will act on — reviews, search results, keyword economics, AI-answer visibility — BedrockTap: one API over 253 sources, a price quoted before each call, a ceiling on the run and a free tier that needs no card. For raw high-volume fetching from sites you have already chosen, Oxylabs, Zyte, ScraperAPI and Decodo are the serious infrastructure options, and Apify is the one to pick when the job needs a scraper nobody has written yet.
BedrockTap’s free tier is 1,000 credits a month, renewed monthly, with no card — about 2,200 Reddit posts or 95 search-results pages. Firecrawl publishes a free plan of 1,000 credits a month; Zyte gives $5 of credit for 30 days and ScraperAPI a seven-day trial of 5,000 credits. Bright Data’s own MCP server carries 5,000 free requests a month.
Not for anything in this catalogue. One HTTPS call in, structured rows out — the fetching, the retries and the shape of each upstream stay behind the API. You never choose a proxy type, a region or a concurrency limit.
Yes. It is a JSON-RPC API over HTTPS: send `Authorization: Bearer <token>` and a JSON body from `curl`, Python, a Lambda or anything else that speaks HTTP. MCP clients get first-class support because that is what agents register, but nothing about the service requires one.
Six of the 253 in the catalogue. Each states what it answers, what it cannot, and which neighbouring source to prefer.
| Source | id | What it gives you |
|---|---|---|
| Google Results & AI Overview | google-serp | Live Google results: organic rankings, the AI Overview with its cited source URLs, People Also Ask questions, and related queries — in one call. |
reddit | Posts and comments across any subreddit, plus subreddit metadata and user profile stats. | |
| Trustpilot Reviews | trustpilot | Consumer and e-commerce business reviews: full text, star rating, reviewer name, date and permalink. |
| Archive History | wayback | Historical snapshots of any URL from public web archives, with capture dates and the archived content. |
| Amazon Products | amazon | Product search, details, offers, best sellers, deals and seller info across Amazon marketplaces. |
| Hacker News | hackernews | Stories, comments, polls and jobs — search plus full item detail. |
By the shape of the task: Web extraction & monitoring · Market research · Competitive intelligence · E-commerce & retail