Perplexity AI MCP Server

chat_completion

The Sonar model searches the web, synthesizes information, and provides a concise answer. This is the basic query tool for factual questions, summaries, and general knowledge. Use this for quick lookups where you need accurate, up-to-date information. Ask Perplexity AI a question and get a grounded, cited answer

chat_with_citations

Each claim or fact in the response is linked to its original source. This is essential for research, fact-checking, and academic work where sources matter. The response includes a citations array with URLs of all referenced sources. Ask Perplexity AI and get answers with source citations

chat_with_domain_filter

Provide domains as a comma-separated list (e.g., "arxiv.org,nih.gov,github.com"). Only sources from the specified domains will be used in generating the answer. Use this for domain-specific research, academic papers, or trusted sources only. Citations are automatically included to verify sources. Ask Perplexity AI restricting search to specific domains

chat_with_history

Provide messages as a JSON array of {role: "user"|"assistant"|"system", content: "text"} objects. This enables follow-up questions where the model understands previous context. Use this for complex queries that build on previous answers or require contextual understanding. Example: [{ "role": "user", "content": "What is quantum computing?" }, { "role": "assistant", "content": "Quantum computing uses quantum bits..." }, { "role": "user", "content": "How does it differ from classical computing?" }] Ask Perplexity AI with multi-turn conversation history

chat_with_images

The response includes an images array with URLs to related images found during the search. Use this for visual topics, product searches, or when you need images to accompany the answer. Ask Perplexity AI and get relevant images with the answer

chat_with_recency_filter

Available recency filters: "hour", "day", "week", "month", "year". This ensures the answer is based on recent information only. Use this for news, recent events, or time-sensitive queries where outdated info is not useful. Ask Perplexity AI with results filtered by time recency

chat_with_related_questions

The response includes a related_questions array with suggested questions for further exploration. Use this for research, learning, and discovering related topics you might want to explore. Ask Perplexity AI and get related follow-up questions

deep_research

This model performs extensive web searches and generates detailed reports with thorough citations. It takes longer than regular queries but provides much more depth and breadth. Use this for complex topics, literature reviews, competitive analysis, or thorough investigations. Maximum tokens default to 4096 for comprehensive responses. Perform deep research with exhaustive web search and comprehensive report

follow_up

Provide the conversation history as a JSON array of messages and the follow-up question. This maintains context from previous turns in the conversation. Use this for multi-turn research sessions where each question builds on previous answers. Ask a follow-up question in an ongoing conversation with Perplexity AI

list_models

Use this to discover what models are available before choosing which one to use for your queries. List all available Perplexity AI models

reasoning

This model excels at multi-step reasoning, mathematical problems, code analysis, and chain-of-thought tasks. Use this for problems requiring step-by-step analysis, mathematical proofs, code reviews, or logical deductions. Citations are included where external information is referenced. Ask Perplexity AI for complex logical reasoning and step-by-step analysis

search_query

This combines all search features: cited sources, relevant images, and follow-up questions. Use this when you want the fullest possible search result with all supplementary information. The response includes content, citations array, images array, and related_questions array. Perform a comprehensive web search with citations, images, and related questions

structured_query

The model will return the answer as JSON matching your schema definition. Provide the JSON schema as a string. This is useful for programmatic data extraction, API integrations, and when you need consistent, parseable responses. Example schema: { "type": "object", "properties": { "name": { "type": "string" }, "age": { "type": "number" } } } Ask Perplexity AI and get a structured JSON response following a schema

system_prompt_query

The system prompt defines how the model should respond (e.g., "You are a medical expert...", "Answer in bullet points..."). Use this for specialized queries, role-playing, formatting requirements, or domain-specific expertise. Example system prompt: "You are a senior software architect. Explain concepts with code examples." Ask Perplexity AI with a custom system prompt to set behavior and context