Perplexity's top tier — when the answer engine is your main tool
The highest Perplexity subscription tier, for people whose research runs through an answer engine all day: the largest allowances for the heavier research and agent modes, priority access to new features, and the fewest limits on long multi-step investigations. → Reach for it only after the normal tier has actually become the constraint — the chapter on Perplexity itself teaches the tool, and most readers never hit the ceiling. → What you are buying is headroom, not a different product; if the deciding factor is EU hosting or company rollout, Langdock above is the more relevant chapter.
Not written yet — this chapter is a placeholder for Perplexity Max, one of the all-in-one AI platforms. Research and copy still to come.
Upgrade only after the standard tier’s limits start hindering your workflow—once you need extra allowance for heavy research, agent modes, or priority feature access.
No; it’s the same answer engine with expanded capacity—think of it as buying extra bandwidth rather than a separate tool, unless EU hosting or company rollout specifics matter.
A saved instruction set that automatically shapes all future Perplexity responses.
Configure Perplexity so every new conversation follows the tone and style you define
The official Python SDK that lets you call Perplexity APIs from code.
You will be able to run Python scripts that authenticate with Perplexity’s services.
pip install perplexityai.config.py.import os followed by os.environ["PERPLEXITY_API_KEY"] = "your_api_key_here" replacing the placeholder with your real key.python -c "import config; print('API key set')" to verify no errors appear.A recurring Perplexity task that runs a research query on schedule and emails you a PDF summary.
Create a hands‑free workflow that delivers a weekly AI news briefing to your inbox
Perplexity’s Search API returns ranked web results as structured JSON.
You will retrieve and display five current web results for a query of your choice.
search.py and import the SDK with from perplexity import Perplexity.client = Perplexity().python
search = client.search.create(
query="latest AI research trends",
max_results=5,
search_context_size="high"
)
``search.results and print each result’s title and URL.python search.py.A live web page with charts and filters that lets users explore side‑by‑side comparisons of AI products.
Generate, host and access a live dashboard that lets you explore AI tools without writing code
Perplexity’s Agent API preset “low” which bundles a model, search config, and tool usage for everyday research.
You will receive a summarized answer with inline citations for a multi‑step question.
agent_low.py and import the SDK as before.client = Perplexity().python
response = client.responses.create(
preset="low",
input="Summarize the core findings of the original 'Attention Is All You Need' transformer paper and explain why it changed NLP."
)
print(response.output_text)
``python agent_low.py.[web:1].[web:1], showing the preset performed web‑grounded reasoning.The Sonar API’s media endpoint that accepts a base64‑encoded image and returns a textual description.
You will encode a local PNG, send it to the API, and receive a descriptive response.
sample.png in your working directory.image_analysis.py and add:
``python
import base64, os
from perplexity import Perplexity
client = Perplexity()
with open("sample.png", "rb") as f:
b64 = base64.b64encode(f.read()).decode("utf-8")
image_uri = f"data:image/png;base64,{b64}"
completion = client.chat.completions.create(
model="sonar-pro",
messages=[{
"role": "user",
"content": [
{"type": "text", "text": "Can you describe this image?"},
{"type": "image_url", "image_url": {"url": image_uri}}
]
}]
)
print(completion.choices[0].message.content)
``python image_analysis.py.sample.png, confirming the API successfully processed the base64 image.67 outcomes in all — one per recipe below.
Perplexity Computer breaks a user‑defined outcome into sub‑tasks and automatically assigns each to the most suitable model from its 19‑model pool, running them in parallel. This coordination lets you get research, code, design, and visuals without switching tools.
Instead of listing step‑by‑step instructions, describe the desired final product and its format; the system figures out the how. Specific deliverable details (PDF, dashboard, charts) guide model selection and output quality.
By asking Perplexity Computer to research a topic and create a structured report with sections, key findings, and citations, it runs parallel research agents and compiles the output into a downloadable PDF.
Specify that you want an interactive dashboard comparing AI tools, including charts and recommendations, and Perplexity Computer will route tasks to research, UI design, and web‑app models, delivering a live page you can share.
When you request a simple web app for tracking subscriptions with inputs, totals, projections, and pie charts, Perplexity Computer generates the code, hosts it, and provides a responsive interface ready to use.
By defining a weekly schedule, source criteria, and output format, Perplexity Computer sets up an autonomous agent that scans new content, summarizes top stories, and delivers a PDF each Monday.
Need answers only from a certain site or date range
Perplexity supports Google‑style operators like site: and after:. By appending these to your query you limit the search to a particular domain or time range, ensuring more relevant and up‑to‑date answers.
I keep retyping the same prompt
Typing '/' opens a menu where you can save custom prompts as shortcuts. These shortcuts auto‑populate the query box with predefined instructions, saving time on repetitive tasks.
Want regular research updates
Perplexity Pro lets you schedule a query to run automatically at set intervals. This creates a hands‑free workflow for regular research or monitoring tasks.
Need answers that rely only on your uploaded docs
Spaces act like private workspaces where you can upload documents, add URLs, and set source toggles. Queries inside a space draw only from its contents (and any enabled sources), letting you ground answers in your own data.
When one model isn’t right for the job
Perplexity Pro lets you choose between models like Claude, Gemini, GPT, or the default ‘best’. Different models excel at research, writing, multimodal input, or logical reasoning; swapping them tailors output quality to the job.
Need a quick fact, an in‑depth report, or a generated spreadsheet
Perplexity offers three modes: Search (fast fact‑finding), Research (multi‑step deep dive with many sources), and Labs (AI agents that generate documents, spreadsheets, etc.). Selecting the right mode balances speed versus thoroughness.
Want AI answers from Chrome’s address bar
By adding Perplexity as a custom search engine in Chrome (or any Chromium browser) and setting it as default, you can invoke AI‑powered answers directly from the address bar, streamlining access.
Keep unpublished research private
Turning off AI data retention prevents Perplexity from storing the content you input, which safeguards sensitive results before publication. This setting gives researchers confidence that their proprietary data stays private.
When I need specialist answers for research
Customizing the occupation, field, and response style tells Perplexity to answer as a specialist, yielding more academic tone, synthesis first, then evidence, then gaps. This improves relevance for research queries.
Keeping literature‑review prompts, PDFs and guidelines together
Spaces act like dedicated workspaces where you can store prompts, uploaded PDFs, guidelines, and custom instructions for a single research task, eliminating repetitive file uploads.
Need a repeatable way to draft abstracts
Skills capture a sequence of prompts and formatting rules that can be invoked repeatedly, ensuring consistent output for tasks such as drafting abstracts or tables.
Need search to pull Wiley articles and Drive docs
Connectors link Perplexity to external repositories, ensuring that search results draw directly from trusted sources like Wiley journals or your own stored documents on Google Drive.
Need a quick literature review draft
The “Computer” feature runs multi‑step, parallel agents that gather sources, synthesize findings, and output a formatted draft, speeding up early‑stage review work.
Need a quick overview of recent clinical studies
Perplexity offers ready‑made workflows (e.g., “Research Update: summarize recent clinical studies”) that automate routine summarization tasks without custom coding.
Want the AI to keep your preferred tone in every chat
The profile tab lets you add custom instructions that apply to every conversation, similar to ChatGPT's custom instructions. This ensures the AI consistently follows your preferred tone, style, or role without re‑entering prompts each time.
Want answers only from academic papers or news feeds
Focus lets you restrict which categories of sources (e.g., academic, news, social) Perplexity searches, improving relevance and trustworthiness for specific queries like scientific data or real‑time news.
Want the same AI role and past answers in every new chat
Collections store a series of related threads plus an overarching instruction (e.g., “act as a travel agent”). New conversations within that collection inherit the role and can reference prior answers, streamlining multi‑step projects.
Want answers from a private PDF you upload
Free users can attach PDF files (including converted images) so Perplexity can cite information directly from those documents, enabling Q&A on private or offline content.
Hitting the free five‑search limit
Pro users get up to 600 searches per day and a richer set of sources, resulting in more detailed answers. This is useful for heavy research tasks where the free five‑search limit would be restrictive.
Need a detailed answer that shows reasoning and cites sources
Perplexity’s search lets you choose models and source filters (e.g., web, social, finance). By selecting the appropriate model and enabling opinion sources, the AI pulls from many cited sites and shows its reasoning steps, giving more accurate, nuanced results than plain chat.
I need a thorough, source‑backed report fast
The Deep Research button runs a timed search that gathers dozens of sources, creates a structured report, and can email you when finished. It’s ideal for thorough investigations with minimal hallucinations.
Have raw research notes and want a web page
Perplexity’s ‘Convert to Page’ feature formats research output with sections, images, and styling, then lets you edit, preview, and publish a shareable link.
When I need the same AI help repeatedly
Spaces act like custom GPTs: you give a system prompt, name the space, and it retains context across sessions, perfect for recurring workflows such as trip planning or stock screening.
Need a daily briefing without any effort
Perplexity’s Task feature lets you set a recurring query, choose model and sources, and define delivery method (email or WhatsApp). It runs automatically at the chosen time.
Need AI answers on WhatsApp
By linking your WhatsApp number, Perplexity can send you task results or answer queries directly in the messaging app, keeping interactions private and mobile‑friendly.
Want to locate exact parts of a YouTube video
Comet, Perplexity’s AI‑native browser, can watch a video page and, using the side assistant, summarize content and locate specific moments, saving time on long media.
Need to pull research results into a Zapier flow
Using the Perplexity API key, you can call the intelligent search endpoint from tools like n8n or Zapier to automate research, summarization, and downstream actions (e.g., email podcast).
Want an offline version of your research page
After creating a page or report, you can export it in multiple formats for offline use or publishing, preserving citations and layout.
Need AI answers that sound like you
In the account settings you can edit the 'Introduce yourself' prompt to define tone, reasoning style, and accuracy preferences. This guides the model to prioritize first‑principles reasoning and ask clarifying questions, reducing hallucinations.
Need a deep research report with tables and sources
Switching the search mode to 'Deep research' tells Perplexity to compile an in‑depth report, pulling from multiple sources and structuring key takeaways, tables, and citations automatically.
Need regular AI answers sent to me
Perplexity lets you create a task that runs on a chosen cadence, using any prompt you define. You can set delivery method (email, notification) and expiration date, turning the AI into a personal monitoring service.
Need answers just from academic papers
You can toggle source types such as Web, Academic papers, Social media, or SEC filings. Restricting sources improves relevance for specific domains and reduces unwanted noise.
Want an easy way to see your research metrics
The Labs feature accepts a prompt describing desired visualizations, style, and data source. Perplexity builds an interactive dashboard that updates when you modify filters or remove outliers.
Want to co‑author a project with teammates
Spaces act like shared workspaces where threads, files, and instructions are stored. You can invite others via a link, allowing real‑time collaboration on itineraries, research, or any project.
Want to see Gmail or Drive info inside a chat
Connectors let you link Gmail, Outlook, Google Drive, Dropbox, or WhatsApp so that queries can fetch live emails, documents, or messages without leaving the platform.
Need a different AI for each query
Perplexity allows you to pick a model per query (Sonar, GPT‑4, Claude, etc.). Selecting a specialized model can improve accuracy for niche domains or reduce cost.
I want to share a whole conversation
After a conversation, you can convert the entire thread into a formatted page with sections, images, and citations. The page can be published via a link for easy sharing.
Need hands‑free answers
The platform includes a microphone icon that enables voice input and reads responses aloud, allowing hands‑free queries and accessibility for users who prefer speaking.
Perplexity automatically searches the web for each query and attaches clickable citations to every fact, letting you verify information instantly. This reduces hallucinations compared to static LLMs.
Need a full slide deck from one prompt
Computer mode treats your request as an AI agent workflow, breaking it into sub‑tasks, selecting optimal models for each step, and assembling slides automatically.
Want current numbers from your apps
Connectors let Perplexity access APIs of platforms such as Google Drive, Notion, YouTube Analytics, etc., so AI can read and act on your personal data securely.
Deep research runs an extended search, aggregates multiple sources, and formats results (including comparison tables) for complex analysis tasks.
Want to keep chats, files and prompts together for repeated work
Spaces act like folders that store related chats, files, and custom prompts, enabling you to reuse instructions across repeated tasks without re‑typing.
Want a search to run automatically every week
The scheduled search feature lets you define a query (or deep‑research/computer workflow) to run at set intervals, delivering results via notification or email.
When enabled, Perplexity sends your question to several top‑tier models, aggregates their responses, and returns a refined answer that reduces single‑model bias.
This mode breaks down a difficult concept into incremental steps, asks you to confirm understanding after each, effectively acting as an interactive tutor.
All files produced by Perplexity (reports, spreadsheets, presentations) are saved in the Artifacts tab, making it easy to retrieve past results without searching chat history.
Need a repeatable way to run the same computer steps
Skills are reusable instruction sets that tell Computer mode how to handle specific types of work (e.g., formatting reports), letting you standardize workflows across projects.
Want the AI to access Google Drive or Notion without API keys
Connectors let the AI agent access your Google Drive, OneDrive, Notion, etc., so it can read/write files, send emails, and manipulate data without manual API keys. The built‑in integrations simplify setup and expand automation possibilities.
The platform stores bio facts, preferences, projects and past conversations in shared memory, allowing later prompts to recall earlier research without re‑entering data. This reduces repetition for long‑term workflows.
Need daily Reddit research
Perplexity Computer can store a prompt as a recurring job that runs at a specific time each day, acting like a cron schedule without external tooling.
The usage panel shows how many credits each operation (text, image, video) consumes, letting you budget against your monthly allowance and avoid surprise overruns.
Need a quick view of Nvidia’s price history
Perplexity Computer can fetch historical price data via public APIs, plot it frame‑by‑frame and export an animated GIF, all from a single natural‑language request.
Need a live view of commercial ship locations
By specifying a public API endpoint, Perplexity Computer can write code that pulls live vessel data, visualizes it on an interactive map, and hosts the app—all without manual coding.
Split your query into three parts—Goal (what you want), Inputs (data sources to use), and Constraints (limits like time range). This forces the model to focus on what matters, yielding more accurate and useful responses.
Different models excel at different tasks: Claude for writing/research, ChatGPT for multi‑step reasoning and coding, Gemini for broad research, Sonar for fast fact retrieval. Picking the right model boosts accuracy and speed.
When I want answers that know my business details
Spaces let you upload private data and set custom instructions, giving Perplexity persistent knowledge about your role, goals, and assets. This turns generic answers into personalized advice.
Want answers just from my Slack, Gmail or Dropbox
Focus mode toggles between source categories (academic, finance, social, etc.) while connectors let Perplexity pull directly from tools like Slack, Gmail, Dropbox. This narrows results to the most relevant domain.
Need a deep analysis with sources and charts
Activating deep research tells Perplexity to gather more sources, create charts, and perform extended reasoning, producing in‑depth analyses like market studies or product launch plans.
Need regular updates like weekly stock summaries
Perplexity’s Tasks feature lets you schedule queries (e.g., weekly stock summaries) that run automatically on a chosen cadence, delivering results via notification.
Need a report, slides or dashboard without formatting
Labs provides a suite of agents that can create structured outputs (docs, slides, dashboards) from prompts. By crafting strong prompts you can produce professional assets without manual formatting.
The same set on /recipes, filtered by tool and role.
The most tangible look at Comet's agentic browsing, the headline perk of Perplexity's premium tiers.
You describe the final product you want, and Perplexity Computer splits that outcome into sub‑tasks. It then picks the best model from its 19‑model pool for each task and runs them in parallel, delivering research, code, design, or visuals all together.
Instead of writing step‑by‑step instructions, you state the desired final deliverable (for example a PDF report or an interactive dashboard) and its format. The system decides how to achieve that, which leads to clearer prompts and more accurate, properly formatted results.
Yes; by defining a schedule, source criteria, and output format, Perplexity sets up an autonomous agent that scans new content each week, summarizes the top stories, and delivers a PDF every Monday.
Open the Profile tab, enter your custom instructions in the self‑introduction field (e.g., “respond like a consultant, use simple language”), and save. Those instructions are then applied automatically to every new conversation.
Use the Focus dropdown on the Home tab (or the Sources filter in the chat toolbar) and select ‘Academic’. The query will then pull information solely from academic publications, improving relevance for scholarly topics.
Perplexity
Perplexity ComputerOutcome‑Focused PromptingPerplexity ProfileFocus FeatureCollectionsAttach PDFsDeep Research ModeScheduled TasksSpaces CollaborationConnectorsModel SelectionVoice & DictationAsk, share, or report — over on the Heidelberg AI community forum.