GPT-5.6 (Sol / Terra / Luna) is now evaluated on TrustVector — with day-1 independent verification, incl. METR's benchmark-cheating findings.

Read the evaluation
Evaluation record · mcp-server-perplexity

MCP Perplexity Server

v2026.7

Perplexity AI

MCPsearchresearchcitationsmcp
84
Strong
About This MCP

Perplexity's official MCP server (perplexityai/modelcontextprotocol, npm @perplexity-ai/mcp-server) for the Perplexity API Platform. Exposes four tools — perplexity_search (Search API), perplexity_ask (sonar-pro), perplexity_research (sonar-deep-research), and perplexity_reason (sonar-reasoning-pro) — providing multi-source research with automatic citations and recency filtering for comprehensive information retrieval.

Last Evaluated: July 9, 2026
Official Website

Trust Vector Analysis

Dimension Breakdown

🚀Performance & Reliability
+
search accuracy

Search quality testing

Evidence
Perplexity API — Advanced semantic search with multi-source aggregation
highVerified: 2026-07-09
citation quality

Citation accuracy testing

Evidence
Perplexity Citations — Automatic inline citations with source verification
highVerified: 2026-07-09
response quality

Response quality assessment

Evidence
Perplexity Responses — Synthesized answers from multiple sources with coherent narratives
highVerified: 2026-07-09
recency filtering

Recency testing

Evidence
Perplexity Features — Configurable time filters for recent information
highVerified: 2026-07-09
api reliability

Reliability testing

Evidence
Perplexity Status — High availability with enterprise-grade infrastructure
highVerified: 2026-07-09
🛡️Security
+
api key security

Authentication review

Evidence
Perplexity Auth — Bearer token authentication with key rotation support
highVerified: 2026-07-09
data handling

Data handling review

Evidence
Perplexity Privacy — Queries processed through Perplexity infrastructure
mediumVerified: 2026-07-09
content safety

Content safety assessment

Evidence
Perplexity Safety — Content moderation and safe search options
mediumVerified: 2026-07-09
enterprise security

Enterprise security review

Evidence
Perplexity Enterprise — SOC 2 compliance and enterprise security features
highVerified: 2026-07-09
🔒Privacy & Compliance
+
query privacy

Privacy policy review

Evidence
Perplexity Privacy Policy — Query data used for service improvement and model training
mediumVerified: 2026-07-09
enterprise privacy

Enterprise privacy review

Evidence
Perplexity Enterprise — Enterprise tier offers data isolation and no training on queries
highVerified: 2026-07-09
data retention

Retention policy review

Evidence
Perplexity Terms — Standard retention policies with deletion options
mediumVerified: 2026-07-09
👁️Trust & Transparency
+
documentation quality

Documentation review

Evidence
Perplexity Docs — Comprehensive API documentation with examples
highVerified: 2026-07-09
source attribution

Attribution testing

Evidence
Perplexity Citations — Excellent source attribution with inline citations
highVerified: 2026-07-09
model transparency

Model transparency review

Evidence
Perplexity Models — Model options documented with capability descriptions
mediumVerified: 2026-07-09
pricing clarity

Pricing review

Evidence
Perplexity API Pricing — Published per-token pricing for all Sonar models: sonar $1/$1 per 1M tokens (in/out), sonar-pro $3/$15, sonar-reasoning-pro $2/$8, sonar-deep-research $2/$8 plus citation ($2/1M), reasoning ($3/1M), and search-query ($5/1K) fees; per-request search fees scale with context size ($5-$12 per 1K requests for sonar, $6-$14 for sonar-pro)
highVerified: 2026-07-09
⚙️Operational Excellence
+
ease of setup

Setup assessment

Evidence
Perplexity Quickstart — Simple API key setup with immediate access
highVerified: 2026-07-09
response speed

Performance testing

Evidence
Perplexity Performance — Fast response times with streaming support
highVerified: 2026-07-09
model selection

Model options review

Evidence
Perplexity API Platform docs — Current Sonar model lineup spans sonar, sonar-pro, sonar-reasoning-pro, and sonar-deep-research, each mapped to a dedicated MCP tool for search, ask, reason, and research workflows
highVerified: 2026-07-09
cost efficiency

Cost analysis

Evidence
Perplexity API Pricing — Base sonar at $1/$1 per 1M tokens is competitive, but per-request search fees ($5-$14 per 1K requests depending on model and context size) add up for high-volume agents; deep-research calls carry additional citation/reasoning/search-query fees
mediumVerified: 2026-07-09
integration quality

Integration testing

Evidence
Perplexity MCP Server guide — Official MCP server (@perplexity-ai/mcp-server v0.9.0) installed via npx with one-click installers for Cursor and VS Code; stdio transport with PERPLEXITY_API_KEY env var
highVerified: 2026-07-09
Strengths
  • +Excellent citation and source attribution
  • +Multi-source aggregation for comprehensive answers
  • +Advanced semantic search capabilities
  • +Recency filtering for current information
  • +Enterprise tier with enhanced privacy
  • +High-quality synthesized responses
Limitations
  • !Queries processed through Perplexity servers
  • !Consumer tier data may be used for training
  • !No self-hosted option for the underlying API
  • !Costs can accumulate with heavy usage — per-request search fees stack on top of token pricing
  • !Rate limits based on pricing tier
  • !Local stdio server only — no hosted remote MCP endpoint documented
Metadata
license: MIT (MCP server); proprietary hosted API service
supported platforms
0: All platforms with Node.js
programming languages
0: TypeScript
mcp version: 1.0
website: https://www.perplexity.ai/
github repo: https://github.com/perplexityai/modelcontextprotocol
package: @perplexity-ai/mcp-server
package version: 0.9.0
api dependency: Perplexity API Platform (Sonar models + Search API)
authentication: Bearer Token (PERPLEXITY_API_KEY)
first release: 2024
maintained by: Perplexity AI
transport types
0: stdio
installation methods
0: npm
1: npx

Use Case Ratings

code generation

Good for finding documentation and coding solutions

customer support

Useful for researching product information

content creation

Excellent for research with automatic citations

data analysis

Helpful for gathering market and trend data

research assistant

Outstanding for academic and professional research

legal compliance

Good for researching regulations; verify with primary sources

healthcare

Can find medical information; not for clinical decisions

financial analysis

Good for market research and financial news

education

Excellent for educational research with citations

creative writing

Great for research and fact-checking