Skip to content
TheChatArchiveThe Chat Archive
Claudeclaude-sonnet-4-20250514· AnthropicHugging Face LMArena Human Preference 140K

Perplexity is not so good?

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Geminigemini-2.5-pro’s answer.
  1. User

    Perplexity is not so good?

  2. Claude Assistant

    Perplexity has some notable strengths but also limitations that might make it less ideal for certain research tasks compared to other AI tools.

    Perplexity's Strengths:

    • Real-time web search integration, so it can access current information beyond training cutoffs
    • Provides source citations for its responses, which is valuable for research verification
    • Good for quick fact-checking and getting up-to-date information
    • User-friendly interface that combines search with AI reasoning

    Perplexity's Limitations for Research:

    • The underlying AI models (typically GPT-3.5 or similar) may not be as sophisticated as GPT-4 or Claude for complex reasoning tasks
    • Search-based responses can sometimes prioritize recent or popular sources over authoritative academic ones
    • Less effective for deep analytical tasks that don't require web search
    • May struggle with nuanced academic concepts that require extensive context understanding
    • Limited customization for specific research domains

    When Perplexity Works Well:

    • Getting current news, statistics, or recent developments
    • Initial research phases when you need broad overviews
    • Fact-checking claims with cited sources
    • Finding recent papers or developments in fast-moving fields

    When Other Tools Might Be Better:

    • Deep analysis of complex academic papers (GPT-4, Claude)
    • Domain-specific research (specialized models like PubMedBERT)
    • Mathematical proofs or theoretical work (specialized tools)
    • Large-scale data analysis (custom ML frameworks)

    Perplexity is a solid tool for certain research tasks, particularly when you need current information with citations, but it's not necessarily the "best" for all research scenarios. The choice really depends on what type of research you're doing and what specific capabilities you need.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.