check_perplexity
What Perplexity answers to a prompt right now, with a live web search from a chosen country, and the sources it cites. ₹0.67 per prompt.
What it does
What Perplexity answers to a prompt right now, with a live web search steered to a country: data.items[] holds the answer as sections[] of Markdown text, each with annotations[] naming the cited sources (title, url), plus data.model_name, token counts and data.fan_out_queries. Use it alongside check_ai_visibility to compare how a brand appears across AI answers ("best protein bar in India"). One live answer per call, so wording varies between runs; it does not search the web for you (search_web).
India: country: "IN" (uppercase ISO code); the default is a worldwide search. The country steers the web search, not the language of the answer: ask for the language in the prompt.
Inputs
Generated from the contract captured from staging on 2026-11-05.
| Field | Type | Required | Description |
|---|---|---|---|
prompt | string | one of | <em>prompt for the AI model</em><br><strong>required field</strong><br>the question or task you want to send to the AI model;<br>you can specify <strong>up to 500 characters</strong> in the <code>user_prompt</code> field. e.g. "provide information on how relevant the amusement park business is in France now" |
country | string | no | <em>country code for web search localization</em><br>optional field<br>specify the country ISO code to get localized web search results<br><strong>Note:</strong> available only for Perplexity Sonar models<br>example: <code>US</code>. e.g. "FR" |
max_output_tokens | integer | no | <em>maximum number of tokens in the AI response</em><br>optional field<br>minimum value: <code>1</code><br>maximum value: <code>4096</code>;<br>default value: <code>2048</code>;<br><strong>Note:</strong> if the reasoning model is specified in the request, the output token count may exceed the specif. e.g. 200 |
One of these sets must be complete (anything else is invalid_input, ₹0): prompt.
What drives the cost
Nothing you can turn: flat per prompt. max_output_tokens caps the length of the answer. Usually five to ten seconds.
Price today
| Price today | ₹0.67 per prompt (cost_paise: 67) at ₹93.93 per US dollar on 17 November 2026 |
| Hold before the call | ₹1.01 (hold_paise: 101); 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 | 90 s, then upstream_timeout (₹0, retryable) |
| Over MCP | run_tool with tool: "check_perplexity" |
| Over REST | POST https://api.parel.si/v1/run with "tool": "check_perplexity" |
| 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":"check_perplexity","input":{"prompt":"best CRM for small businesses in India","country":"IN"}}'
What comes back
data with items[]. A real answer recorded during the live check of 2026-10-28, trimmed:
{
"model_name": "sonar",
"input_tokens": 14,
"output_tokens": 486,
"reasoning_tokens": 0,
"web_search": true,
"datetime": "2026-10-28 11:29:45 +00:00",
"items": [
{
"type": "message",
"sections": [
{
"type": "text",
"text": "For most small businesses in India, **Zoho Bigin** is the best starting choice: it is simple, pipeline-focused, and priced from about **₹400/user/month** on annual billing.[1][3]\n\n| CRM | Best for | Indicative starting price |\n|---|---|---:|\n| **Zoho Bigin** | Small teams wanting a simple sales pipeline | From ₹400/user/month; free plan available[1][3] |\n| *…",
"annotations": [
{
"title": "Best CRM for Small Business in India",
"url": "https://www.salesforce.com/in/crm/crm-for-small-business/best-crm/?bc=OTH",
"start_index": null,
"end_index": null,
"text": null
},
{
"title": "CRM For Small Business (Your Complete Guide)",
"url": "https://www.salesforce.com/in/crm/crm-for-small-business/?bc=OTH",
"start_index": null,
"end_index": null,
"text": null
},
"… 8 more item(s) truncated"
]
}
]
}
],
"fan_out_queries": null
}
Notes
- Usually five to ten seconds.
- Answers are not deterministic: run a prompt two or three times before concluding a brand is absent.
- India: country "IN". A country steers the web search, not the language of the answer; ask for the language in the prompt.
Siblings
check_ai_visibilityfor ChatGPT's answer to the same prompt.search_webfor what Google shows.
This page as Markdown: /docs/tools/check_perplexity.md. Something wrong or missing: hello@parel.si.