screenshot_page
A real-browser screenshot of any web page as a PNG link on Parel's own address, kept for 7 days. ₹0.27 per screenshot; a page that cannot be captured is ₹0.
What it does
A real-browser screenshot of one web page from its url. Parel stores the picture and hands back a link under its own address: data.image is the PNG link (about 80 KB for a plain page), data.url the address that was captured and data.image_expires_at when the copy is removed, 7 days after the call. The answer is a link, not the image itself, so it stays small for any client, including MCP; fetch the link to view, save or show the picture, and save the file if it must outlive the 7 days. viewportWidth sets the browser width in pixels (default 1280).
Use it to check how a page looks (a landing page, a competitor's offer, a layout) rather than what it says: read_page gives the text and extract_structured the data. It captures the viewport as first painted, so an animation or a late banner can be missing; there is no full-page or scrolling option and it cannot pass a login.
India: no locale knob; the page is captured as it is served to a visitor.
Inputs
Generated from the contract captured from staging on 2026-11-05.
| Field | Type | Required | Description |
|---|---|---|---|
url | string | yes | The full URL of the page to capture.. e.g. "https://example.com" |
viewportWidth | integer | no | Browser viewport width in pixels (e.g. 1280).. default 1280 |
What drives the cost
Nothing: flat per screenshot. A page that cannot be captured is class: "ok_free", ₹0. A picture that could not be stored on Parel's side is a free upstream_error (image_fetch_failed or similar) that is safe to retry. Each account has a daily allowance of stored pictures (50 MB a day by default): past it the call is refused before anything is captured, ₹0, with daily_picture_limit and resets_at (midnight India time). Every other tool still works. Usually four to eight seconds, up to twenty.
Price today
| Price today | ₹0.27 per screenshot (cost_paise: 27) at ₹93.93 per US dollar on 17 November 2026 |
| Hold before the call | ₹0.41 (hold_paise: 41); when the call settles the real price is charged and the unused part is released |
| ₹0 when | it fails, or the provider answers but there is nothing in it (an empty list, no email, no page): class: "ok_free", ₹0 |
| Timeout | 30 s, then upstream_timeout (₹0, retryable) |
| Over MCP | run_tool with tool: "screenshot_page" |
| Over REST | POST https://api.parel.si/v1/run with "tool": "screenshot_page" |
| Confirmation | not needed (the estimate is under the ₹200 threshold) |
Request
curl https://api.parel.si/v1/run \
-H "Authorization: Bearer $PAREL_KEY" \
-H "Content-Type: application/json" \
-d '{"tool":"screenshot_page","input":{"url":"https://example.com"}}'
What comes back
data.image, data.url and data.image_expires_at. A real answer recorded during the live check of 2026-11-05 (example.com in 3.8 s, a 77 KB PNG stored and served from Parel's own address; this particular link has since expired):
{
"image": "https://api.parel.si/i/7f268fcc769f514a7e42e6723b0c464a.png",
"url": "https://example.com/",
"image_expires_at": "2026-12-04T19:34:18.031Z"
}
The link is public and needs no key: anyone holding it can open the picture until it expires, so share it as you would any screenshot. It is cacheable for a day and never changes.
Notes
- Flat per screenshot. data.image is a PNG link on Parel's own address, not the image bytes; the copy is kept for 7 days (data.image_expires_at), then the link answers 404. Save the file if it must outlive that. Past the account's daily allowance of stored pictures the call is refused, free, with daily_picture_limit and resets_at (midnight India time).
- Captures the first paint of the viewport (1280 px wide by default): animated or late-loading content may be missing; no scroll, no full-page option, no login.
- Usually four to eight seconds, up to twenty. A page that cannot be captured, or a picture that could not be stored, is free and says so (error image_fetch_failed or similar, retryable).
Siblings
read_pagefor what the page says.extract_structuredfor the page as a table.
This page as Markdown: /docs/tools/screenshot_page.md. Something wrong or missing: hello@parel.si.