Perplexity AIreview, pricing and limits

Last updated: 2026-07-01

Perplexity AI is an AI answer engine that searches the live web for every query, not a chatbot: it pairs results with numbered source citations and now drives over $450 million in yearly recurring revenue. Deep Research extends that layer into harder, multi-page questions. Live retrieval is what separates it from tools drawing only on a static training snapshot.

About Perplexity AI

Perplexity AI is an AI-powered answer engine built by Perplexity AI, Inc. (founded August 2022), designed to search the live web and return structured, cited answers in seconds. As of early 2026, it serves 45 million monthly users and processes over 1 billion queries per month, generating $450 million in annualized recurring revenue. Unlike static chatbots that rely on frozen training data, Perplexity retrieves real-time information from indexed web pages and pairs each answer with numbered source links, making it the primary tool for research that requires verifiability. The platform runs a multi-model architecture coordinating up to 19 distinct AI models, anchored by its proprietary Sonar engine, itself built on top of Llama 3.3 70B with extra fine-tuning for search relevance and factual accuracy. That Cerebras-hosted inference keeps answers near-instant even under complex queries. For deeper research tasks, Pro and Max users get Deep Research mode, which crawls a large batch of web sources per query, checks them against each other for contradictions, and turns the results into a multi-section report with full citations in 2 to 5 minutes, using Claude Opus 4.5 and 4.6 to reason over the material. In 2026, Perplexity expanded from an answer engine into a full agentic platform. Perplexity Computer, launched February 25, 2026, is an autonomous AI agent that executes multi-step workflows by browsing the web, operating software, filling forms, and spawning sub-agents for parallel tasks. Model Council, launched February 5, 2026, runs a single query across three AI models simultaneously so users can compare outputs side by side. The Comet browser embeds Perplexity search into an AI-native browser with granular privacy controls and autonomous task execution for actions like booking flights and managing email. The platform runs on web, iOS, Android, and macOS, with the Comet browser extending the same search and agent capabilities directly into everyday browsing. Pricing scales from a permanent free tier up to Enterprise Max, detailed below. Perplexity's clearest advantage over ChatGPT is citation accuracy: in head-to-head tests, Perplexity returns more verifiable sources with fewer hallucinated URLs. Its main weakness is creative tasks and deep code debugging, where ChatGPT and Claude remain stronger. Backed by $1.72 billion across 11 funding rounds from Jeff Bezos, Nvidia, and Databricks, Perplexity reached a $22.6 billion valuation by January 2026.

Screenshots

Pricing

Free tier with unlimited basic searches and 5 Pro Searches/day. Pro at $20/month (or $200/year). Max at $200/month (or $2,000/year). Education Pro at $10/month for verified students. Enterprise Pro at $40/user/month; Enterprise Max at $325/user/month. Sonar API starts at $1/million input tokens and $1/million output tokens.

Key Features

• Real-Time Web Search with Citations: Every answer pulls from live web sources indexed in real time, returning numbered citations so users can verify each claim in under 10 seconds.
• Deep Research Mode: Autonomously visits 100 or more web pages per query, cross-references data, flags contradictions, and generates a structured multi-section research report in 2 to 5 minutes. Deep Research is now available inside Perplexity Computer, letting users turn findings directly into reports, spreadsheets, decks, or follow-up workflows in the same session.
• Multi-Model Architecture: Coordinates up to 19 distinct AI models and routes each query to the optimal model; Model Council lets users run one query through three models simultaneously to compare outputs. Max subscribers can now select the model powering their Comet browser agent, with Opus 4.6 as the default.
• Comet Browser (Enterprise and Consumer): Perplexity Comet browser is free on iOS, Android, Windows, and Mac with a built-in AI assistant for research, meetings, coding, and task automation, reportedly increasing user question volume by 6-18x. Now available for Enterprise: admins can deploy Comet silently across macOS and Windows devices via MDM, configure browser policies, and control which actions the AI agent can take. Also integrates into Microsoft Word, Excel, PowerPoint, and Outlook.
• Perplexity Computer and Personal Computer: Perplexity Computer (launched February 2026) is an autonomous AI agent that browses the web, operates software, fills forms, and creates parallel sub-agents. Voice Mode (June 2026) lets users describe tasks, give feedback, or redirect mid-task by voice inside Computer. Personal Computer (June 2026) is an always-on AI running on a dedicated Mac mini that monitors triggers, executes proactive tasks, and carries work forward 24/7 by merging local files, apps, and cloud sessions. Personal Computer for Windows, announced June 3 2026, extends the same orchestration to Windows with native access to Excel, PowerPoint, Word, Outlook, and OneDrive. June 2026 also added Deep Research in Computer (visits 100+ sources and converts findings into reports, spreadsheets, or dashboards), a Command Panel (slash-accessed discovery interface for Deep Research, Plan Mode, and custom skills), Forking (branch a thread into a fresh follow-up while keeping full prior context), and a Computer Analytics API for enterprise admins to track per-member credit usage.
• Sonar API: OpenAI-compatible search API starting at $1 per million tokens, letting developers add real-time, citation-grounded AI search to any application with a single API call.
• Multimodal File and Image Analysis: All plans support image uploads up to 50 MB and file analysis; Enterprise Pro unlocks files up to 1 GB with 1-year retention, plus video transcription added November 2025.
• Hybrid Local-Server Inference Orchestrator: Rolling out to Perplexity Computer on Windows in July 2026, a compact local model routes subtasks by data sensitivity: financial records and health files stay on-device while compute-heavy tasks route automatically to frontier cloud models, with no manual configuration.

Pros

• Returns verifiable citations for every answer: in direct tests, Perplexity sourced 8 peer-reviewed URLs with zero hallucinated links vs ChatGPT returning 4 with 1 fabricated.
• Free tier includes unlimited basic searches with no account required, making it accessible with zero upfront cost vs competitors like Kagi that charge from $5/month.
• Sonar runs at 1,200 tokens per second on Cerebras infrastructure, matching or exceeding frontier model speed at a fraction of the compute cost for search-grounded tasks.

Cons

• Weak creative writing and code debugging: users who need original content generation or complex code review consistently rate ChatGPT and Claude as significantly better.
• Trustpilot reviews average 1.6/5 from 180+ users, dominated by billing complaints about automatic subscription renewals and difficulty cancelling Pro plans.
• Citations are minimal format (numbered inline), making the output unsuitable for academic publishing or formal reports that require MLA, APA, or Chicago style references.

Visit Perplexity AI Official Website