Perplexity Review
The research-first AI engine — an answer engine that cites its sources, now with a Comet browser that's free on every platform, Perplexity Computer orchestrating frontier models (Sonar 2, GPT-5.6 Terra, Gemini 3.7 Flash and Claude Sonnet 5.5 among them), and access to GPT-5.6, Claude and Gemini under one subscription.
At a Glance
Overview
Perplexity is the odd one out in the AI assistant landscape — and that's exactly its strength. It's not trying to be your all-purpose chatbot, your coding partner, or your creative writing muse. Perplexity is an answer engine: ask a question, get a cited, sourced, verifiable answer with links to where the information came from. In an era where AI confidently hallucinates, Perplexity's commitment to grounding every answer in real sources makes it uniquely trustworthy for research, fact-checking, and knowledge work.
As of September 2026, Perplexity has evolved well beyond its search-engine roots. Perplexity Computer — an agentic system that orchestrates 19 different AI models to execute complex multi-step workflows — and the Comet browser (a full AI-native web browser that acts as a personal assistant) have transformed Perplexity from a research tool into a platform. Comet is now free on iOS, Android, Windows and Mac, having launched as a $200/month PC-only subscription, with Comet Plus at $5/month (included with Pro and Max) unlocking premium publisher content. Pro subscribers get the best models from OpenAI (GPT-5.6 Terra), Anthropic (Claude Sonnet 5.5), and Google (Gemini 3.7 Flash) alongside Perplexity's own Sonar 2 — all through Perplexity's interface, with citations and source tracking built in.
Perplexity isn't the best AI for writing, coding, or creative work. But for anyone who needs to know things — researchers, journalists, analysts, students, and the insatiably curious — it's the most trustworthy AI assistant available. And at $20/month for Pro, or $200/year with a limited-time $17/month annual rate, it gives you models that individually cost $20/month each — arguably the best value in AI.
Key Features
- Cited, sourced answers — Every response includes inline citations with clickable links to source material. Perplexity's core differentiator: you can verify every claim. No other AI assistant makes fact-checking this easy.
- Perplexity Computer — This agentic system orchestrates 19 different AI models to execute complex, multi-step workflows autonomously on research, code, documents and connected tools. Research a topic, compile a report, analyze data, and create presentations — all in one session. It requires a subscription and consumes credits: Pro includes 4,000 bonus credits, Max 35,000. Computer also runs inside Microsoft 365 apps (Word, Excel, PowerPoint, Outlook, Teams), and the June 19 changelog added Deep Research integration inside Computer plus faster command access and forking of long-running tasks.
- Comet browser — Perplexity's own AI-native web browser. Acts as a personal assistant that can automate tasks, research the web, organize your email, and interact with websites on your behalf. It's now free on all platforms — iOS, Android, Windows and Mac — after launching as a $200/month PC-only subscription. Comet Plus ($5/month, included with Pro and Max) unlocks premium publisher content. Worth knowing: Perplexity has said publicly that Comet is partly designed to collect data for ad targeting.
- Multi-model access — Pro and Max subscribers can choose which model answers their query, from a roster that now includes Sonar 2, GPT-5.6 Terra, Gemini 3.7 Flash, Claude Sonnet 5.5, Kimi K3, GLM 5.3, Grok 4.6 and Nemotron 3 Ultra — plus GPT-5.6 Sol and Claude Opus 5.5 on Max. One subscription, all the major models.
- Deep Research — Multi-step research that searches dozens of sources, synthesizes findings, and produces cited reports. 20 queries/day on Pro. Comparable to ChatGPT and Gemini's Deep Research but with Perplexity's citation advantage.
- Pro Search — Unlimited on Pro plan. Goes beyond simple keyword matching to understand complex questions, search multiple sources, and synthesize nuanced answers with context.
- Sonar models — Perplexity's own in-house AI models, led by Sonar 2, optimized specifically for search-grounded, citation-backed responses. Sonar Pro is available via API for developers at $3 input / $15 output per 1M tokens.
- Model Council and Learn Mode — Model Council runs multiple frontier models against the same question and compares their answers side by side, while Learn Mode turns Perplexity into a guided tutor. Learn Mode is included with Education Pro.
- Spaces and Collections — Collaborative research workspaces where teams can share queries, build knowledge bases, and organize research threads. Collections let you save and categorize your research history.
- File upload with citations — Upload PDFs, documents, and spreadsheets. Perplexity analyzes them and answers questions with citations pointing to specific sections of your uploaded files.
- Labs — Experimental features and early access to new capabilities. Perplexity ships fast and Labs is where you try things before they're polished.
Pricing
Perplexity's pricing is straightforward and competitive, with the Pro plan offering exceptional value when you consider it includes access to models that individually cost $20/month.
- Perplexity Standard (Free) — Basic search with very limited Pro Searches per day and no advanced models or image generation. Good for casual fact-checking and light research. You'll hit the search limits quickly if you use it as a daily driver.
- Perplexity Pro ($20/month or $200/year, limited-time $17/month annual) — The sweet spot and best value in AI. Unlimited Pro Search, 20 Deep Research queries/day, expanded Perplexity Computer access with 4,000 bonus credits, file upload with citations, Create files and apps, and access to GPT-5.6 Terra, Gemini 3.7 Flash, Claude Sonnet 5.5 and Sonar 2. The annual rate works out cheaper than any individual model subscription.
- Perplexity Max ($200/month, around $167/month billed annually) — Everything in Pro with dramatically higher limits: maximum Computer usage, 35,000 bonus credits, expert-level research, and frontier models including GPT-5.6 Sol and Claude Opus 5.5. For power researchers, journalists, and analysts who use Perplexity as their primary work tool all day.
- Perplexity Education Pro ($10/month) — Pro features plus Learn Mode, for verified students and educators.
- Enterprise Pro ($40/seat/month, or $400/year) — Team workspaces, admin controls, an org file repository, shared Spaces and Collections, usage analytics, and SSO. For organizations standardizing on Perplexity for research. An Enterprise Pro for Government tier runs $0.25/agency for the first 15 months under the GSA OneGov program.
- Enterprise Max ($325/seat/month, or $3,250/year) — Enterprise Pro plus the trust center, strict data privacy guarantees, and priority support. For regulated teams that need Perplexity under tighter controls.
- Comet Plus ($5/month) — Premium publisher content inside the Comet browser. Included with Pro and Max.
- API (Sonar models) — Pay-per-token access to Perplexity's Sonar models for developers building search-grounded AI applications; Sonar Pro runs $3 input / $15 output per 1M tokens.
For most individual users, Pro at $20/month is an easy recommendation — it's the same price as ChatGPT Plus and Claude Pro, but gives you access to all the major labs' models plus Perplexity's own research tools. The value proposition is unmatched, and verified students can get the same toolkit through Education Pro at $10/month.
Pros
- Every answer is cited with clickable sources — uniquely trustworthy for research and fact-checking
- Pro at $20/month includes GPT-5.6 Terra, Gemini 3.7 Flash, Claude Sonnet 5.5 and Sonar 2 access — best value in AI
- Perplexity Computer orchestrates frontier models for autonomous multi-step workflows on subscription credits
- Comet browser is a genuinely useful AI-native web experience that automates browsing tasks — and it's now free on every platform
- Sonar models are optimized for search-grounded responses — less hallucination than general-purpose models
- Deep Research produces well-cited, synthesis-heavy reports comparable to ChatGPT and Gemini
- Spaces and Collections make collaborative research and knowledge management easy
- API access to Sonar models at $3/$15 per 1M tokens for developers
Cons
- Not a general-purpose assistant — poor at creative writing, weak at coding, no image generation
- Free tier limits are aggressive — you'll hit Pro Search caps quickly with regular use
- Max at $200/month is steep for what's essentially higher usage limits, not new capabilities
- Multi-model access means slower responses — routing through Perplexity adds latency vs using models directly
- Comet browser is still maturing — fewer extensions, some sites break, and the AI can be overeager
- No voice mode or real-time audio — text-only interface
- Sonar models are good but not best-in-class — GPT-5.6 and Claude Opus 5.5 outperform them on pure reasoning
- Comet's free model comes with a data trade-off — Perplexity has said the browser is partly designed to collect data for ad targeting
Use Cases
Verdict
Perplexity occupies a unique and valuable niche in the AI landscape. It's not trying to be ChatGPT or Claude — it's an answer engine built for people who need to know things, verify claims, and trust their sources. The commitment to citing every answer makes it the most trustworthy AI assistant available for research and knowledge work.
The Pro plan at $20/month is arguably the best deal in AI right now: unlimited Pro Search, 20 Deep Research queries/day, and access to GPT-5.6 Terra, Claude Sonnet 5.5, Gemini 3.7 Flash and Sonar 2 through one interface. If you currently pay for multiple AI subscriptions, Perplexity Pro could replace two or three of them for research-focused work.
The limitations are real: Perplexity is not your writing assistant, your coding partner, or your creative collaborator. It can't generate images. Its Sonar models are solid but not best-in-class for pure reasoning. And the free tier limits are tight enough that serious users will need to pay.
For researchers, journalists, analysts, students, and anyone who values verifiable answers over confident-sounding prose — Perplexity is an 8/10 and the best tool for the job. For general-purpose AI assistance, ChatGPT and Claude remain stronger. But if you care about knowing whether what your AI tells you is actually true, Perplexity is essential.
Frequently Asked Questions
Is Perplexity free?
Yes, the Standard tier gives you basic search with very limited Pro Searches per day and no advanced models or image generation. Perplexity Pro costs $20/month (or $200/year, with a limited-time $17/month annual rate) and adds unlimited Pro Search, Deep Research, 4,000 bonus Computer credits, and access to GPT-5.6 Terra, Claude Sonnet 5.5, Gemini 3.7 Flash and Sonar 2. Max is $200/month for much higher limits, and Education Pro is $10/month for verified students.
How is Perplexity different from ChatGPT?
Perplexity is an answer engine — every response includes cited sources with clickable links. ChatGPT is a general-purpose AI assistant. Perplexity excels at research, fact-checking, and verifiable knowledge work. ChatGPT excels at writing, coding, creative tasks, and has broader capabilities (image generation, voice, plugins). They're complementary tools, not direct competitors.
What is Perplexity Computer?
Perplexity Computer is an agentic AI system that orchestrates 19 different AI models to execute complex, multi-step workflows autonomously. It can research topics, compile reports, analyze data, and create presentations in a single session. It requires a subscription and consumes credits — Pro includes 4,000 bonus credits and Max includes 35,000. Computer also runs inside Microsoft 365 (Word, Excel, PowerPoint, Outlook, Teams). The June 19 update added Deep Research inside Computer, plus faster command access and the ability to fork long-running workflows.
Does Perplexity Pro include ChatGPT and Claude?
Yes. Perplexity Pro ($20/month) gives you GPT-5.6 Terra (OpenAI), Claude Sonnet 5.5 (Anthropic), Gemini 3.7 Flash (Google) and Kimi K3, GLM 5.3, Grok 4.6 and Nemotron 3 Ultra through Perplexity's interface — all with citations and source tracking — and Max adds GPT-5.6 Sol and Claude Opus 5.5. You can choose which model answers each query. It's the only subscription that includes all the major AI labs' models.
What are Perplexity's Sonar models?
Sonar 2 is Perplexity's latest in-house model, optimized specifically for search-grounded, citation-backed responses. It's designed to minimize hallucination and maximize source accuracy. Sonar Pro is sold separately on the API at $3 input / $15 output per 1M tokens. While not as capable at pure reasoning as GPT-5.6 or Claude Opus 5.5, Sonar is excellent for research tasks.
Can Perplexity generate images or code?
Perplexity can generate basic code through its integrated models but is not a strong coding assistant — ChatGPT Codex and Claude Code are far better. Perplexity does not natively generate images, though you can access image generation indirectly through the integrated GPT-5.6 model on Pro. If image generation or coding are your primary needs, ChatGPT or Claude are better choices.
Is the Comet browser free?
Yes — Comet is now free on all platforms (iOS, Android, Windows and Mac). It launched as a $200/month PC-only subscription before Perplexity dropped the paywall. Comet Plus at $5/month, included with Pro and Max, adds premium publisher content. Worth knowing: Perplexity has said publicly that Comet is partly designed to collect data for ad targeting.