---
title: "Perplexity AI Review 2026: Features, Pricing & Promo Code"
description: "In-depth Perplexity AI review covering Pro features, pricing, Deep Research, AI models, and a $10 promo code. Is the AI answer engine worth it in 2026?"
date: 2026-07-26
categories: ["ai"]
tags: ["perplexity","ai-search","answer-engine"]
---
import Button from "@components/widgets/Button.astro";
import Notice from "@components/widgets/Notice.astro";
import ListCheck from "@components/widgets/ListCheck.astro";
import Accordion from "@components/widgets/Accordion.astro";
import Tabs from "@components/widgets/Tabs.astro";
import Tab from "@components/widgets/Tab.astro";
import YouTubeEmbed from "../../components/widgets/YouTubeEmbed.astro";
I've been using [Perplexity.ai](https://go.bitdoze.com/perplexity) daily for over two years. In that time I've canceled ChatGPT Plus, paused Claude Pro, and let Gemini Advanced lapse. Perplexity is the one AI subscription I keep paying. This review explains why.
Since I first wrote about it in early 2024, the platform has changed massively. It's grown from ~15 million monthly active users to over 45 million MAU, processing 780 million queries per month. The company raised $1.72 billion total and hit a $22.6 billion valuation. This isn't a niche tool anymore. It's a serious contender against Google for informational search.
The competitive landscape has shifted too. ChatGPT, Claude, Gemini, and DeepSeek all have search capabilities now. Perplexity needs to earn its place, and this review covers where it still does and where it doesn't.
The original version of this article covered Perplexity's basic free and Pro tiers. Since then, the platform has added: **Deep Research** (reports in minutes from 100+ sources), a **$200/month Max tier**, the **Comet AI browser**, **Shopping/Buy with Pro**, **Spaces/Projects** for collaboration, and a full **Sonar API** for developers. Scroll to each section for details.
## What is Perplexity? The AI answer engine changing search
Perplexity is an AI answer engine, not a chatbot and not a traditional search engine. You ask a question, it searches the web in real time, reads multiple sources, and synthesizes a cited answer. Every claim links back to its source.
This is completely different from how Google works (a list of 10 blue links you have to click through) and how ChatGPT works (a conversational response that may or may not have web grounding). Perplexity gives you the answer with inline citations so you can verify every claim.
The pipeline: **Your question, Perplexity searches the web, reads N sources, synthesizes a cited answer**. Simple concept, but execution quality is what keeps people paying.
The name comes from a language model metric: "perplexity" measures how well a model predicts text. Lower perplexity means better predictions. A fitting name for a product that makes AI predictions more grounded in reality.
Perplexity crawls websites to build its answers. In October 2024, News Corp (Dow Jones, NY Post) sued Perplexity for "massive illegal copying" of copyrighted content. Perplexity lost its bid to dismiss the case in August 2025. This is an ongoing legal risk worth knowing about. If you run a website and want to control which AI systems can access your content, see our guide on how to [block AI crawlers from scraping your site](/block-ai-crawlers).
## Perplexity free vs Pro: which plan is right for you?
The free tier is genuinely useful. You can evaluate Perplexity without paying a cent and decide if it fits your workflow.
### Perplexity free tier: what you get
Free users in 2026 get:
- **Unlimited basic (Sonar) searches**: the default fast model handles straightforward questions
- **3 Pro Searches per day**: multi-step reasoning with tool use (searching, reading, synthesizing)
- **Limited Deep Research access**: enough to try it, not enough for heavy use
- **Basic focus modes**: narrow search to web, academic sources, or specific sites
Three Pro Searches per day is enough to evaluate whether the feature justifies paying. If you find yourself hitting that ceiling, that's your signal to upgrade.
### Perplexity Pro features: is the upgrade worth it?
Pro at $20/month unlocks:
- **Unlimited Pro Search** (with weekly fair-use limits, more on this below)
- **All AI models**: GPT-5.2, Claude Sonnet 4.6, Gemini 3.1 Pro, Nemotron 3 Super 120B, plus the full Sonar family
- **Image generation** with multiple engines
- **File uploads**: analyze PDFs, images, documents
- **$5/month API credits** (reported by users (see the API section for details))
- **Priority access during high traffic**
The cost argument is straightforward: Pro at $20/month gives you access to GPT-5.2 AND Claude Sonnet AND Gemini AND search-grounded answers. Subscribing to ChatGPT Plus ($20) and Claude Pro ($20) separately costs double and you still don't get Perplexity's citation-grounded search.
If you're a verified student or educator, Education Pro is $10/month with the same features via SheerID verification.
## Perplexity Pro Search: multi-step reasoning with citations
The old "Copilot" branding is gone. What used to be called Copilot mode is now **Pro Search**: and it's better.
Pro Search does multi-step reasoning: it breaks down complex questions, searches multiple times, reads different sources, asks clarifying follow-up questions when needed, and produces a comprehensive cited answer. It's the feature that separates Perplexity from "just another AI chatbot with a search button."
**How limits work in practice:**
| Tier | Pro Search limit |
|---|---|
| Free | 3 per day |
| Pro | Weekly limits ("average use", exact number not published) |
| Max | Unlimited |
When you hit the Pro Search limit, the answer still comes back, but it uses basic Sonar search instead of multi-step reasoning. Not a hard block, which is better than hitting a wall.
My practical advice: save Pro Search for complex research questions. Basic Sonar search handles simple factual queries fine.
## AI models available on Perplexity (2026)
Perplexity runs a dual-layer model system. Understanding this is key to getting the most from the platform.
### Sonar family: Perplexity's proprietary models
The Sonar models are Perplexity's own, built on Llama and optimized for speed and citation grounding:
- **Sonar**: the default for most queries. Fast, grounded, good enough for 80% of searches
- **Sonar Pro**: more capable, handles complex queries
- **Sonar Reasoning Pro**: multi-step reasoning for harder problems
- **Sonar Deep Research**: dedicated model for the Deep Research feature (see below)
Sonar is what you'll use most. It's fast, the citations are reliable, and for factual questions it's hard to beat.
### Third-party models: GPT-5.2, Claude, Gemini and more
Pro and Max users can switch between third-party models:
| Model | Tier | Best for |
|---|---|---|
| GPT-5.2 (OpenAI) | Pro | General reasoning, coding |
| Claude Sonnet 4.6 (Anthropic) | Pro | Long-form writing, analysis |
| Claude Opus 4.6 (Anthropic) | Max | Complex reasoning, agentic tasks |
| Gemini 3.1 Pro (Google) | Pro | Multimodal, long context |
| Nemotron 3 Super 120B (NVIDIA) | Pro | Concise, fast responses |
**Which model for which task:** Use Sonar for factual questions with citations. Switch to GPT-5.2 or Claude Sonnet when you need deeper reasoning or longer-form writing. Gemini 3.1 Pro shines on multimodal queries (image analysis). Claude Opus on Max is for the hardest problems.
The Sonar family is based on Llama. If you're interested in [open-source alternatives to Claude and GPT](/best-open-source-llms-claude-alternative), the Sonar models are a good example of what fine-tuned Llama can do at scale.
**Writing Mode** still exists. It lets you chat directly with your chosen model without web search. Be aware: Writing Mode disables search entirely. Answers come from training data only, with no citations. It's useful for creative writing or brainstorming, but don't expect sourced answers.
## Perplexity Deep Research: reports in minutes
The biggest addition since this article was first published. Launched in February 2025, Deep Research performs dozens of searches, reads hundreds of sources, and produces a comprehensive report in 2-4 minutes.
**Benchmark performance:** 21.1% on Humanity's Last Exam, 93.9% on SimpleQA. Those scores matter because they show the model can handle both hard reasoning and factual accuracy.
**Access by tier:**
- Free: limited access (enough to try it)
- Pro: generous allocation for regular use
- Max: unlimited
**What it's good for:**
- Technology comparisons and stack decisions
- Market analysis for a new project
- Learning a new domain quickly
- Troubleshooting complex technical problems where you'd normally have 20 browser tabs open
**What it's not good for:** vague prompts. "Tell me about AI" produces a generic report. "Compare Docker vs Podman for self-hosting a PostgreSQL cluster on ARM with 4GB RAM" produces something genuinely useful. Be specific.
Use Deep Research for technology evaluations and stack decisions. It reads 50+ sources and produces structured reports with citations. If you want to build your own multi-agent research pipeline instead, check out our guide on building a [multi-agent research team with Google ADK](/google-adk-multi-agent-search).
## Image generation on Perplexity
Image generation is a Pro feature. The current engines (as of mid-2026):
- **GPT Image 1** (OpenAI): strong general-purpose generation
- **Seedream 4.5** (ByteDance): good for stylized and artistic images
- **Nano Banana** (Google): fast, lightweight generation
The image generation is convenient when you're already in a Perplexity conversation and need a quick illustration. For fine control, custom models, or batch generation, dedicated tools are still better. If you want to [generate AI images locally on Mac](/ai-images-mac) with more control, Flux and ComfyUI give you that.
## Perplexity Max: the $200/month power tier
Launched in July 2025, Max is Perplexity's premium tier at $200/month. It includes:
- **Unlimited Labs usage**: early experimental features
- **Early access to new products** (Comet browser launched to Max first)
- **Claude Opus 4.6 + o3-pro**: models not available on Pro
- **Brain memory system**: Perplexity remembers context across conversations
- **Priority support**
Max is worth it if you need Claude Opus for complex reasoning, unlimited Deep Research, or you spend 8+ hours/day in Perplexity. For 90% of individual users, Pro at $20/month covers everything. If you're a student or educator, Education Pro at $10/month is the better value.
## Comet browser and Perplexity Computer: AI agents beyond search
Perplexity is expanding beyond search with two major products.
### Comet browser
Perplexity's own AI-first browser, built on Chromium. Comet Assistant can see web pages you're browsing, summarize emails, manage tabs, navigate websites, and execute multi-step workflows. It's an AI agent that lives in your browser.
Comet launched in July 2025 to Max subscribers first, then a waitlist. It's ambitious. Perplexity even made an unsolicited $34.5 billion bid to buy Google Chrome, capitalizing on the DOJ antitrust ruling. That bid probably won't succeed, but it shows where Perplexity sees itself heading.
### Perplexity Computer
An agentic AI system that coordinates multiple frontier models to complete multi-step desktop tasks: booking travel, filling forms, running research workflows. This is Perplexity's move into AI agents beyond search.
**Honest take:** Comet is more immediately useful. It's a real browser you can use today. Perplexity Computer is more ambitious but less proven. Both are evolving fast.
If you'd rather build custom AI agents yourself, you can [build your own AI agent with Mastra](/build-ai-agent-mastra). For raw web data access, [AI-powered web scraping with BrightData](/brightdata-mcp-guide) ([Bright Data](https://go.bitdoze.com/brightdata)) gives you structured data at scale.
## Perplexity Chrome extension and security warning
The current official extension is **"Perplexity - AI Search"** on the Chrome Web Store.
What it does:
- Highlight text on any webpage and ask Perplexity about it
- Quick search from any tab
- Get definitions and explanations inline while browsing
In June 2026, Microsoft discovered a **fake malicious Perplexity Chrome extension** that intercepted users' searches and exfiltrated data. Only install the verified extension from the official Chrome Web Store. Double-check the publisher name and install count before adding any browser extension that handles your search queries.
If you run a website, note that Perplexity crawls sites to build its answers. You may want to [block AI crawlers from scraping your site](/block-ai-crawlers) if you prefer to control access.
## Perplexity API: Sonar, Agent and Search APIs
The API has evolved far beyond the "$5 monthly credits" model from 2024. Perplexity now offers four distinct APIs:
**Sonar API**: search-grounded LLM responses. This is the API most developers will use.
| Model | Input (per 1M tokens) | Output (per 1M tokens) |
|---|---|---|
| Sonar | $1 | $1 |
| Sonar Pro | $3 | $15 |
| Sonar Deep Research | $2 | $8 |
**Agent API**: access to third-party models (OpenAI, Anthropic, Google, xAI) at provider rates with no markup.
**Search API**: raw web search results at $5 per 1,000 requests.
**Embeddings API**: for semantic search/RAG, $0.004-$0.05 per 1M tokens.
Official Python and TypeScript SDKs are available. There's also an MCP Server for Claude Desktop, Cursor, and VS Code.
If you're running your own AI infrastructure, the Sonar API and Search API can replace building your own web scraping pipeline. You get search-grounded responses via a clean API without managing crawling infrastructure. [TinyFish](https://go.bitdoze.com/tinyfish) is another option if you need web search, fetch, and browser APIs for AI agents.
```bash
curl --request POST \
--url https://api.perplexity.ai/chat/completions \
--header 'accept: application/json' \
--header 'authorization: Bearer $PERPLEXITY_API_KEY' \
--header 'content-type: application/json' \
--data '{
"model": "sonar-pro",
"messages": [{"role": "user", "content": "What are the latest developments in AI search?"}],
"stream": false,
"web_search_options": {"search_context_size": "medium"}
}'
```
```bash
pip install perplexityai
```
```python
from perplexity_ai import Perplexity
client = Perplexity(api_key="your-api-key")
response = client.chat.completions.create(
model="sonar-pro",
messages=[{"role": "user", "content": "How do I set up Docker on Ubuntu 24.04?"}],
)
print(response.choices[0].message.content)
```
Pro subscribers reportedly get $5/month in API credits, though the official docs now describe a pay-as-you-go model. If you're new to APIs, our [getting started with AI programming](/ai-programming-beginners-guide) guide covers the basics.
## Perplexity Spaces, Projects and Shopping
Two features that didn't exist when this article was first published.
### Spaces and Projects
Customizable AI-powered collaboration hubs. Upload files, set custom AI instructions, invite collaborators, and search across your internal files plus the web. Enterprise users get app connectors for Google Drive, SharePoint, and similar services.
**File limits:** Pro gets 50 files per space, Max gets 500, Enterprise Max gets 5,000.
This is useful for teams doing research together. Everyone shares the same Perplexity context with uploaded docs and custom instructions.
### Shopping and Buy with Pro
Perplexity can search for products, compare them with AI recommendations, and complete purchases via "Buy with Pro" one-click checkout. Includes "Snap to Shop" visual search and Shopify integration. Free shipping on Buy with Pro orders.
**Caveat:** Buy with Pro is US-only. If you're outside the US, the product search and comparison features still work, but you can't check out through Perplexity.
## Perplexity pricing and $10 promo code
Here's the full pricing as of mid-2026:
| Plan | Price | Key features |
|---|---|---|
| Free | $0 | Unlimited basic search, 3 Pro Searches/day, limited Deep Research |
| Pro | $20/month | Unlimited Pro Search (weekly limits), all models, image gen, file uploads |
| Education Pro | $10/month | Pro features for verified students/educators (SheerID) |
| Max | $200/month | Everything in Pro + Claude Opus, unlimited Labs, Comet, Brain, priority support |
| Enterprise Pro | $40/seat/month | Team features, admin controls, app connectors |
| Enterprise Max | Custom | Everything + 5,000 files/space, SSO, dedicated support |
**The $10 promo:** You can get your first month of Pro for $10 instead of $20 using the referral link below. You get $10 off, I get $10 off.
## Perplexity vs ChatGPT, Gemini and Claude: why it still matters
Every major AI tool now has search. ChatGPT searches the web. Claude has web search. Gemini is built on Google's index. DeepSeek has search. So why pay for Perplexity?
**Where Perplexity wins:**
1. **Citation transparency**: every claim is sourced with inline citations. ChatGPT and Gemini are improving, but Perplexity's citation density and source quality remain the best I've used.
2. **Multi-model access**: one $20/month subscription gives you GPT-5.2 + Claude + Gemini + Sonar. Subscribing to each separately costs $60+/month.
3. **Research-first design**: Pro Search and Deep Research are purpose-built for research workflows, not bolted-on features.
4. **API value**: the Sonar API and Agent API at provider rates are unique in the market.
**Where Perplexity loses:**
ChatGPT is better for general creative tasks, coding assistance with canvas, and when you need strong general reasoning without citations. Its plugin ecosystem is also more mature.
Claude is better for long documents, nuanced writing, deep analysis of uploaded files, and when you need careful, methodical reasoning. If you're also exploring [AI coding tools and assistants](/ai-coading-tools), Claude's coding capabilities are excellent.
Gemini is better for multimodal tasks (image and video analysis), Google Workspace integration, and very long context windows (up to 2M tokens). If you live in Google's ecosystem, Gemini is the natural fit.
For coding-specific work, dedicated tools often beat general-purpose AI. See our guide to [AI coding tools and assistants](/ai-coading-tools) for a deeper comparison.
## Operator's notes: what I've learned using Perplexity daily
Practical observations from over two years of daily use:
**What's great:**
- Citation quality is consistently high. I can verify claims quickly.
- Deep Research saves hours of manual tab management for tech evaluations
- Switching models mid-conversation is genuinely useful (start with Sonar, switch to Claude for writing)
- The Chrome extension is handy for quick context while reading articles
**What's annoying:**
- Pro Search limits are vague. "Weekly limits for average use" means you don't know where the ceiling is until you hit it
- Image generation quality varies wildly between engines
- Writing Mode confuses new users who expect citations but get training-data-only responses
**What breaks:**
- Occasional hallucinated citations. The source exists but doesn't actually say what Perplexity claims it does. Always click through and verify critical claims
- Shopping only works in the US
- Comet browser is still early and has bugs
**My daily Perplexity workflow:**
Use basic Sonar search for quick factual questions
Switch to Pro Search for multi-source research
Use Deep Research for technology comparisons and stack decisions
Switch to Claude Sonnet when I need long-form writing help
Save Pro Search queries for complex problems. Don't waste them on simple lookups
Always click through citations on anything I'm going to act on or publish
**Rate limit tip:** when you hit Pro Search limits, basic Sonar search still works fine for straightforward queries. Save Pro Search for the complex stuff.
**Cost tip:** if you're paying for both ChatGPT Plus ($20) and Claude Pro ($20), switching to Perplexity Pro alone saves $20/month and you still get access to both models through Perplexity's interface.
## Conclusion: is Perplexity worth it in 2026?
Yes, especially Pro at $20/month.
The core value proposition has only gotten stronger: citation-grounded AI search with multi-model access in one subscription. Deep Research alone justifies the subscription for anyone who does regular research. The fact that you get GPT-5.2, Claude, Gemini, and Sonar for $20/month (when subscribing to each separately costs $60+) makes it the best value in AI subscriptions right now.
The competitive landscape has changed. ChatGPT, Claude, and Gemini all have search now. But Perplexity's citation density and research-first design still set it apart for anyone who needs sourced, verifiable answers rather than conversational responses.
Max at $200/month is only for power users who need Claude Opus or unlimited Deep Research. For most people, Pro is the sweet spot.
The News Corp copyright lawsuit is worth watching. It could reshape how AI search engines interact with publishers. But it hasn't slowed Perplexity's growth so far.
If you're also exploring AI for development, check out our guide to [AI coding tools and assistants](/ai-coading-tools).
Pro at $20/month is the sweet spot. You get GPT-5.2, Claude, Gemini, Deep Research, and citation-grounded search in one subscription. If you're paying for multiple AI subscriptions, consolidating to Perplexity Pro alone saves money.