Perplexity

Editor's Review

Perplexity is the research tool that chats — always-on cited web answers, and in 2026 it pushed deeper into agentic territory. Pricing: Free (~5 Pro searches/day, no Deep Research/Labs); Pro $20/month ($200/year, currently $17/month billed annually as a limited-time flag) with unlimited Pro Search, 20 Deep Research/day, 50 Labs/month, model picker (GPT-5, Claude Opus 4.6), and $5/month Sonar API credits; Max $200/month ($2,000/year) adding 10,000 Computer credits/month, unlimited Labs, and Model Council; Education Pro $10/month. The 2026 features: Model Council (Max-only, February) sends one question to 3 frontier models and synthesizes where they agree and disagree; Perplexity Computer (agentic browsing and tasks) reached Pro subscribers in March after being Max-exclusive; the Comet browser went free for everyone in October 2025; Sonar API runs from ~$1/M tokens. The honest black mark: a major 2026 backlash when Pro limits were reduced mid-contract for annual subscribers — no grandfathering, no refunds. That broke trust in a way pricing pages can't fix. Max at 10× Pro only pays off if you genuinely delegate recurring work to background agents. Who should pay? Researchers and analysts who live in cited answers — Deep Research at 20/day on Pro is the workhorse tier. Who shouldn't? Anyone wanting an open-ended assistant: ChatGPT is the generalist that can search; Perplexity is the search tool that chats. Perplexity wins cited research; ChatGPT wins writing, coding, and brainstorming. And after the mid-contract cuts, pay monthly until trust is re-earned.