Tools Overview

Extend model capabilities with built-in tools and custom functions in the Agentic Research API.

Overview

The Agentic Research API provides tools that extend model capabilities beyond their training data. Tools must be explicitly configured in your API request—once enabled, models autonomously decide when to use them based on your instructions.

web_search

Perform web searches with filtering by domain, language, recency, and date range.

fetch_url

Fetch and extract content from specific URLs.

Function Calling

Define custom functions to connect models to your own systems and APIs.

Built-in vs Custom Tools

Type	Tools	Use Case
Built-in	web_search, fetch_url	Real-time web information retrieval
Custom	Your functions	Connect to databases, APIs, business logic

Quick Example

Enable tools by adding them to the tools array in your request:

Python

from perplexity import Perplexity

client = Perplexity()

response = client.responses.create(
    model="openai/gpt-5.2",
    input="What are the latest AI developments?",
    tools=[
        {"type": "web_search"},
        {"type": "fetch_url"}
    ],
    instructions="Use web_search for current information. Use fetch_url when you need full article content."
)

print(response.output_text)

Pricing

Tool	Cost
web_search	$5.00 per 1,000 calls
fetch_url	$0.50 per 1,000 calls
Function Calling	No additional cost (standard token pricing)

You’re also charged for tokens consumed when tool results are embedded in the model’s context. See the Pricing page for full details.

Next Steps

Web Search

Learn about search filters and localization options.

Fetch URL

Extract content from specific web pages.

Function Calling

Define custom functions for external integrations.

Presets

Use pre-configured presets with built-in tools.