Overview

The Perplexity API Platform MCP Server is the official Model Context Protocol implementation from Perplexity AI, enabling AI assistants to access real-time web search, advanced reasoning, and comprehensive research capabilities through Perplexity's powerful Sonar models and Search API.

Official Implementation:

Developed and maintained by Perplexity AI

Key Features

🔍

Real-Time Web Search

Direct access to current web information with ranked results and metadata. Learn more about Knowledge Graph Memory.

🧠

Advanced Reasoning

Leverage sonar-reasoning-pro for complex problem-solving and analysis. Explore advanced reasoning with Ollama Deep Researcher.

📚

Deep Research

Conduct comprehensive multi-source research with detailed reports. Discover Sequential Thinking for enhanced research.

💬

Conversational AI

Natural language interaction with real-time web context

Available Tools

Quick Reference

Tool	Purpose	Model	Best For
`perplexity_search`	Direct web search	Search API	Finding current information
`perplexity_ask`	Conversational AI with search	sonar-pro	Quick questions & answers
`perplexity_research`	Deep comprehensive research	sonar-deep-research	Detailed analysis & reports
`perplexity_reason`	Advanced reasoning	sonar-reasoning-pro	Complex problem-solving

Detailed Usage

perplexity_search▶

Performs direct web search and returns ranked results with metadata. Ideal for discovering current information and getting quick search results.

use_mcp_tool({
  server_name: "perplexity",
  tool_name: "perplexity_search",
  arguments: {
    query: "latest developments in quantum computing 2025"
  }
});

Returns: Ranked search results with URLs, titles, snippets, and relevance scores

perplexity_ask▶

General-purpose conversational AI with real-time web search using the sonar-pro model. Perfect for getting quick, accurate answers to questions.

use_mcp_tool({
  server_name: "perplexity",
  tool_name: "perplexity_ask",
  arguments: {
    query: "What are the key features of Next.js 15?",
    temperature: 0.7
  }
});

Model: sonar-pro
Returns: Conversational response with web-sourced information

perplexity_research▶

Extended Timeout:

Deep research may take longer. Consider setting PERPLEXITY_TIMEOUT_MS higher (e.g., 600000 for 10 minutes)

Deep, comprehensive research using the sonar-deep-research model. Ideal for thorough analysis and detailed reports on complex topics.

use_mcp_tool({
  server_name: "perplexity",
  tool_name: "perplexity_research",
  arguments: {
    query: "Comprehensive analysis of AI safety measures in 2025",
    temperature: 0.3
  }
});

Model: sonar-deep-research
Returns: Detailed research report with multiple sources and comprehensive analysis

perplexity_reason▶

Advanced reasoning and problem-solving using the sonar-reasoning-pro model. Best for complex analytical tasks and multi-step reasoning.

use_mcp_tool({
  server_name: "perplexity",
  tool_name: "perplexity_reason",
  arguments: {
    query: "Analyze the trade-offs between microservices and monolithic architecture for a startup",
    temperature: 0.2
  }
});

Model: sonar-reasoning-pro
Returns: Detailed logical analysis with step-by-step reasoning

Installation

Prerequisites

API Key Required:

You need a Perplexity API key. Get one from the Perplexity API Portal

{
  "mcpServers": {
    "perplexity": {
      "command": "npx",
      "args": [
        "-y",
        "@perplexity-ai/mcp-server"
      ],
      "env": {
        "PERPLEXITY_API_KEY": "your-api-key-here",
        "PERPLEXITY_TIMEOUT_MS": "300000"
      }
    }
  }
}

Environment Variables:

PERPLEXITY_API_KEY - Your API key from Perplexity (required)
PERPLEXITY_TIMEOUT_MS - Request timeout in milliseconds (optional, default: 300000 = 5 minutes)

Sonar Models

Model Comparison

Model	Purpose	Tool	Speed
`sonar-pro`	General conversational AI with search	perplexity_ask	⚡ Fast
`sonar-reasoning-pro`	Advanced reasoning & problem-solving	perplexity_reason	⚡⚡ Medium
`sonar-deep-research`	Comprehensive multi-source research	perplexity_research	⚡⚡⚡ Thorough

Use Cases

Quick Information Discovery

Use perplexity_search to find current information and get ranked search results for:

Breaking news and current events
Latest product releases or updates
Real-time data and statistics
Recent research papers or articles

Conversational Assistance

Use perplexity_ask with sonar-pro for:

Quick questions with up-to-date answers
Fact-checking and verification
Getting summaries of current topics
General knowledge queries with web context

Complex Problem-Solving

Use perplexity_reason with sonar-reasoning-pro for:

Analyzing trade-offs and decision-making
Technical architecture comparisons
Multi-step logical reasoning
Systematic problem breakdown

Deep Research & Analysis

Use perplexity_research with sonar-deep-research for:

Comprehensive market research
Academic literature reviews
Industry trend analysis
Detailed technical documentation
Multi-source fact compilation

Configuration Tips

Timeout Settings

For different use cases, adjust the timeout:

{
  "env": {
    "PERPLEXITY_TIMEOUT_MS": "300000"  // 5 minutes - default
  }
}

Recommended timeouts:

Search & Ask: 300000 (5 minutes) - default is fine
Reasoning: 450000 (7.5 minutes) - for complex analysis
Deep Research: 600000+ (10+ minutes) - allow time for comprehensive research

Temperature Settings

Control response creativity:

0.0-0.3: Factual, precise, deterministic
0.4-0.7: Balanced creativity and accuracy
0.8-1.0: More creative, varied responses

Troubleshooting

Common Issues

API Key Not Working

Verify your API key is correct from the Perplexity API Portal
Ensure the environment variable is properly set: PERPLEXITY_API_KEY
Check that your API key has sufficient credits
Restart your MCP client after updating the configuration

Timeout Errors on Research Queries

Deep research queries can take longer to complete. Increase the timeout:

"PERPLEXITY_TIMEOUT_MS": "600000" // 10 minutes

Installation Issues

Ensure you have Node.js installed (version 16+)
Try running manually: npx @perplexity-ai/mcp-server
Clear npm cache: npm cache clean --force
Check your internet connection

Poor Search Results

Make queries more specific and detailed
Include relevant time frames (e.g., "2025", "recent", "latest")
Use perplexity_research for comprehensive coverage
Try adjusting temperature settings for different result styles

Perplexity AI MCP Server: Real-time Search & Reasoning for AI