Perplexity Agent API Deep Research Preset

Pre-configured deep research agent using Perplexity Agent API with GPT-5.2, up to 10 reasoning steps, 50-80+ source consultation, web search and URL fetching tools, and 10K max output tokens for institutional-grade analysis.

Overview

The deep-research preset is a fully configured agent within Perplexity's Agent API, optimized for comprehensive multi-source analysis. It uses GPT-5.2 and performs iterative research across up to 10 steps.

Configuration

• Model: openai/gpt-5.2
• Max Output Tokens: 10,000
• Max Tokens Per Page: 4,000
• Max Steps: 10
• Tools: web_search, fetch_url
• Prompt Token Count: ~3,267

Capabilities

• Iterative research with 50-80+ sources for complex queries
• Cross-validation across multiple independent sources
• Domain-expert level writing standards
• Length calibration based on query complexity
• Primary source prioritization
• Inline citations with [1][2] format

Usage

from perplexity import Perplexity

client = Perplexity()
response = client.responses.create(
    preset="deep-research",
    input="Comprehensive analysis of quantum computing market in 2026"
)
print(response.output_text)

Override Options

Override any preset parameter by passing additional parameters:

• model, max_steps, tools, max_output_tokens
• Custom instructions alongside preset defaults