Screenshot API for developers, AI agents and automations

Website screenshot API. Get back a link you can pass on.

One HTTP call renders the page in a real browser and returns a hosted image link. Hand it to your agent, drop it into a Google Sheet, or send it to a vision model. There is no browser for you to run.

  • 150 free screenshots a month
  • No credit card
  • Failed screenshots are free
GET /v1/screenshot 200 OK · 1 credit
$ curl "https://api.renderscreenshot.com/v1/screenshot\
    ?url=https://linear.app/pricing&full_page=true&block_cookies=true" \
  -H "Authorization: Bearer rs_live_••••" \
  -H "Accept: application/json"
{
  "status": "completed",
  "image": {
    "url": "https://cdn.renderscreenshot.com/a8/9f2c41.png",
    "width": 1280, "height": 5120, "format": "png"
  },
  "cache": { "hit": false, "expires_at": "2026-10-16T09:00:00Z" },
  "usage": { "credits": 1, "remaining": 149 }
}

Pass the link to

  • An agent tool result
  • A Google Sheets cell
  • An Airtable attachment
  • A GPT-4o or Claude vision step
  • A Slack message

Where does your screenshot go next?

Taking the screenshot is the easy part. Getting the image to the next step is where most setups break. RenderScreenshot gives every capture a link, so the next step only needs a URL.

For AI agent builders

Keep pixels out of your context window

Your tool gets a URL back, not a base64 blob. The image costs no tokens until the model, or a person, opens it.

// tool call from the model
{ "name": "screenshot_url",
  "input": { "url": "https://staging.acme.dev/checkout", "full_page": true } }

// tool result your code returns
{ "type": "tool_result",
  "content": "https://cdn.renderscreenshot.com/c1/77e0a2.jpg (1280x3400)" }
  • One stateless call. No browser sessions to manage and no browser-hour meter.
  • Short tool inputs: url, full_page, preset. Your code sets everything else, including cookies for logged-in pages.
  • Errors return a code and a retryable flag, so the agent knows whether to try again. Failed renders use no credits.
  • Works as a plain tool in Claude, the OpenAI Agents SDK, the Vercel AI SDK and LangChain.

For automation builders

From a sheet of URLs to a column of screenshots

Use the HTTP step that Zapier, Make and n8n already have. There is nothing to install, so it also works on n8n Cloud.

A · Website B · Screenshot C · Credits
2 acme.com cdn…/a1f0.png 1
3 northwind.io cdn…/b72c.png 1
4 contoso.shop timed out, retry 0
  • You get an image link, not a binary file. It goes straight into a cell, an Airtable attachment or a Slack message.
  • Hide cookie banners and ads, wait for slow pages, and capture the full page with one setting each.
  • 1 screenshot = 1 credit. Failed screenshots cost nothing, and a replayed step can reuse the last screenshot for free.
  • Send the link to an OpenAI or Claude step for website audits and personalized outreach.

Try it on a real page

Pick a site and a size. You get the screenshot and the link the API would return.

Preset:

Screenshot

Pick a site, then take a screenshot

Link you get back

https://cdn.renderscreenshot.com/…
Cost
1 credit
Link lifetime
24 hours by default
Cookie banners, ads
Hidden
If it fails
0 credits

You pay for screenshots that work

Agent loops and long workflows retry a lot. The bill should not grow with every retry, and it should never go past your plan price.

Every format and size costs the same: PNG, JPEG, WebP or PDF, one screen or the full page.

Full credit rules

A new screenshot

Any size, any format.

1 credit

A failed screenshot

Timeouts, errors and rejected requests.

0 credits

A repeat from cache

Same request with cache_available=true.

0 credits

Opening the image link

Served by the CDN, not the API.

0 credits

Your monthly limit

Requests stop until the reset or an upgrade.

No overage bill

Handles the pages that break home-made scripts

Each option is one setting on the same call. You do not need Puppeteer, Chrome or a Docker image.

Cookie banners and ads hidden
Consent pop-ups and ads are removed before the capture.
block_cookies=true
Full page or one element
Capture the whole scroll, or only a chart, table or form.
full_page=true
Slow pages
Wait for the network to go quiet, for a selector, or for a fixed delay.
wait.until
Pages behind a login
Send cookies, headers or HTTP auth with the request.
network.cookies
PDF as well as images
PNG, JPEG, WebP or a print-ready PDF.
format=pdf
Many URLs at once
Up to 100 URLs in one batch request, with a webhook when they finish.
POST /v1/batch
Your own storage
Paid plans can copy every screenshot to an S3-compatible bucket you own.
storage
Keys stay private
Signed URLs let a page or a tool load screenshots without your secret key.
signed URLs

Pricing

One credit is one screenshot. Every plan has every rendering feature. Plans are billed monthly.

Free

$0 /month

150 screenshots a month

About 5 a day

  • 10 requests a minute
  • Batches of up to 5 URLs
  • 2 screenshots at the same time
  • Up to 10 workspaces

Starter

$10 /month

2,500 screenshots a month

About 83 a day

  • 60 requests a minute
  • Batches of up to 20 URLs
  • 5 screenshots at the same time
  • Unlimited workspaces
Most popular

Growth

$40 /month

12,000 screenshots a month

About 400 a day

  • 120 requests a minute
  • Batches of up to 50 URLs
  • 10 screenshots at the same time
  • Unlimited workspaces

Scale

$100 /month

60,000 screenshots a month

About 2,000 a day

  • 200 requests a minute
  • Batches of up to 100 URLs
  • 20 screenshots at the same time
  • Unlimited workspaces

No credit card for the free plan. Need more than 60,000 a month? Email [email protected]. Compare plans in detail.

Questions

Do I get an image file or a link?

Both. Send Accept: application/json and the API returns JSON with image.url, a hosted link that needs no API key. Leave the header out and you get the image bytes, with the same link in the X-Cache-URL header.

How long does the image link work?

For as long as the screenshot is cached: 24 hours by default, or anything from 1 hour to 30 days with cache_ttl. To keep a copy for longer, save it in your workflow (Google Drive, Airtable, S3) or, on paid plans, send every screenshot to your own S3-compatible bucket.

Do I need to install a Zapier, Make or n8n app?

No. Use the HTTP step your tool already has: Webhooks by Zapier, the HTTP module in Make, or the HTTP Request node in n8n. Because nothing is installed, it works on n8n Cloud as well as self-hosted n8n.

Is there an MCP server?

Not yet. Your agent calls the HTTP API through a normal tool definition, which works with Claude, the OpenAI Agents SDK, the Vercel AI SDK and LangChain. The AI agents page has copy-paste tool definitions.

What does a failed screenshot cost?

Nothing. A credit is used only when a render succeeds. Errors return a code and a retryable flag, so an agent or a workflow can decide whether to try again.

Can it capture pages behind a login?

Yes. Pass cookies, headers or HTTP auth with the request. In an agent, set these values in your tool code so the model never sees or chooses them.

What happens when a site blocks automated browsers?

We do not solve CAPTCHAs or disguise the browser. Add a content check so that a challenge page fails the request instead of returning a picture of it, and failed requests use no credits. For sites you own, you can let the renderer through.

Your first screenshot link is one call away

150 free screenshots every month. No credit card.