AgentHubAgentHub

AI tools·Search & research·Global

Perplexity

Cited AI search. Built for research, comparison, and news—not closed-door writing.

9.6/ 10.0Editorial Rating
Editorial score based on public benchmarks and community feedback — not a third-party benchmark result

Compiled through 2026-09

Perplexity visual
Overseas accessFree tierCited searchRealtime infoCompare
Vendor
Perplexity AI
Started
2022
China availability
Overseas accessRequires overseas network access
Pricing tier
Free tier
Current flagship
Perplexity Pro / Deep Research
Website
www.perplexity.ai

Evaluation, Pros & Cons

Editorial score based on public benchmarks and community feedback — not a third-party benchmark result

Key Advantages

  • Gold standard AI search engine; live web retrieval with precise cited references
  • Pro Search & Deep Research multi-step reasoning with switchable frontier backends

Limitations & Caveats

  • Subject to source site indexing and paywalled/anti-scraping limitations

Overview

Perplexity productized search + model summary + citations to fight uncited hallucination. Pro can switch GPT, Claude, Gemini, Grok. Sonar is the search-grounded API. 2025–2026 added deeper Research reports, shopping, and enterprise search. It competes with Gemini and ChatGPT Search as an answer engine, with a stronger citation habit.

Direction

The bet remains verifiable answers: better citations, research agents, browsing/shopping, and Sonar for developers. It will likely keep routing to other labs’ flagships rather than training the largest base model.

Version history

  1. Jul 2026

    Latest

    GPT-5.6 / Claude 5 routing

    Pro picker tracks new flagships.

    View details

    New capabilities

    • Pro picker tracks GPT-5.6 / Claude 5-class models

    What got stronger

    • The search product can summarize with each lab’s newest flagship immediately

    Version notes

    The July 2026 picker update is the strategy: someone else’s model, Perplexity’s UX. Swapping models is routine; citations and Research are the moat.

    #2026-07-gpt-5-6-claude-5-routing

  2. Feb 2025

    Deep Research

    Multi-step research reports.

    View details

    New capabilities

    • Deep Research multi-step reports

    What got stronger

    • Can split a question, read many pages, and write a sectioned report

    Version notes

    Deep Research pulled Perplexity from Q&A into “do an hour of research for me.” You still spot-check citations; the shape is an agent.

    #2025-02-deep-research

  3. Mar 2024

    Sonar API

    Search-grounded API for developers.

    View details

    New capabilities

    • Sonar search-grounded API

    What got stronger

    • Builders could embed cited answers in their own products

    Version notes

    Sonar is Perplexity for developers: not the chat UI, a retrieval-grounded generation API. Fits as the retrieval layer of an internal assistant.

    #2024-03-sonar-api

  4. Nov 2023

    Perplexity Pro

    Paid tier and better models.

    View details

    New capabilities

    • Pro: paid tier, better models, higher limits

    What got stronger

    • Heavy research could be daily, not rationed

    Version notes

    Pro funded better underlying models. The product difference stayed citations and search, not training the largest base model.

    #2023-11-perplexity-pro

  5. Dec 2022

    Perplexity launch

    Conversational search with cites.

    View details

    New capabilities

    • Conversational search with clickable citations

    What got stronger

    • Answers link back to pages—less unsourced invention than plain chat

    Version notes

    Perplexity launched with citations as the anti-hallucination feature. The model was ordinary; the shape was right: a search engine, not a novelist.

    #2022-12-perplexity-launch