Answer engine and Pro Search
Perplexity established conversational web answers, citations, follow-up context and model choice for paid users.
- Real-time web retrieval
- Inline citations
- Conversational follow-ups
- Pro model selector
← AI Tools directory
AI SearchResearch dossier
Perplexity
An answer engine and agent platform built around live retrieval, citations, multi-model reasoning and finished work products.
Perplexity is a retrieval-and-agent product, not one fixed foundation model. It routes or lets paid users choose models, then adds search, citations, code execution and product-specific orchestration.
First released2022
CompanyPerplexity AI
Best forSource-cited web research, multi-model search and deep research
PricingFree; Pro and Max; Max $200/mo or $2,000/year
What Perplexity is today
Perplexity's core workflow interprets a question, searches the live web, synthesizes an answer and attaches citations that users can inspect.
The platform has expanded from search into Research, editable asset creation, the Comet browser and Computer agents that can work across files, apps and the web.
Reading note
Product features and underlying models are documented separately. Roadmap items include only provider-announced work; rumors are excluded.
Major versions and what changed
Selected milestones that materially changed the product.
Deep Research and Labs
Research automated many searches and source reads; Labs added code, charts and richer project outputs.
- Multi-step research
- Hundreds of source reads
- Code execution
- Reports, dashboards and simple apps
Comet AI browser
Perplexity brought an assistant into the browsing layer for page-aware research and actions.
- Browser-context assistance
- Tab and page workflows
- Assistant-led browsing
- Later desktop and mobile expansion
Computer and Advanced Deep Research
Perplexity expanded into longer autonomous work and upgraded Research with documents, calculation, broader browsing and a live progress interface.
- Work across web, apps and files
- Improved code sandbox
- Clarifying and mid-run follow-up questions
- Editable, shareable research reports
Earlier Deep Research vs Advanced Deep Research
The comparison uses provider-published capabilities and named evaluations, not a generic marketing score.
| Dimension | Previous | Current | Measured difference |
|---|---|---|---|
| Research loop | Autonomous searches and a final report | More sources, cross-checking and harder-to-reach web content | More depth and professional due-diligence focus |
| User control | Submit and wait for the report | Clarifying questions and follow-ups while research is running | The user can steer scope before and during execution |
| Tools | Primarily web research | Uploaded documents plus improved calculation/code sandbox | Can mix source review with quantitative work |
| Output | Generated report | Streaming, editable and shareable file | Research moves directly into a working deliverable |
No version-wide multiplier is published for Perplexity.
Perplexity combines changing third-party models with its own retrieval and agent layers. A single 'x more powerful' number would mix model quality, search coverage, latency and tools; the company describes quality improvements but does not publish a universal previous-versus-current multiplier.
Typical Research report
Perplexity's help center says Research can perform dozens of searches and read hundreds of sources in a few minutes.
Check evidenceDRACO benchmark
Perplexity's 2026 research benchmark covers complex deep-research tasks across 10 domains and 40 countries.
Check evidenceOverall improvement
No official cross-version score supports a defensible universal multiplier.
How the important features work
What happens behind the button, where it helps and what to verify.
Search and citations
Core free and paid experienceHow it works: Retrieves live pages, ranks relevant passages, uses a language model to synthesize them and links claims back to sources.
Best use: Current facts, discovery and quick source trails
Watch: Citation presence does not prove that every claim is supported; open the underlying page.
Advanced Deep Research
Compute and model limits vary by subscriptionHow it works: Repeatedly plans, searches, reads, cross-checks and can calculate or analyze uploads before writing an editable report.
Best use: Due diligence, academic exploration and large comparisons
Create files and apps
Eligible Pro, Education, Max and Enterprise plansHow it works: Combines research, code execution, visual generation and iterative prompts to create documents, spreadsheets, slides and HTML apps.
Best use: Turning research into a deliverable
Watch: Documents can include citations, but generated apps and calculations still require functional review.
Computer and Comet
Platform and plan rollout variesHow it works: Uses browser/app context and agent tools to carry work across pages, local or connected files and supported services.
Best use: Longer multi-app workflows
Watch: Review action scope, credentials and any irreversible step before authorizing it.
LIMITS
What the headline features do not guarantee
- Search quality depends on index coverage, source accessibility and ranking; paywalls or blocked pages can leave gaps.
- The selected foundation model may change, so identical prompts can behave differently over time.
- Agent actions and generated assets require human review, especially when financial, legal or account changes are involved.
Announced roadmap, without speculation
Status reflects the provider's latest public wording. “No public date” means there is no verified launch schedule.
Rolling out
Computer platform expansion
Perplexity is extending Computer across desktop, teams, professional verticals and builder APIs.
Official announcementRolling out
Agent API and managed runtimes
The August 2026 Agent API release continues Perplexity's move from consumer search to developer-controlled agents.
Official announcementNo public date
Next named search generation
Perplexity has not announced a single numbered successor because the product evolves continuously across models and modes.
Primary sources
Provider announcements, help documentation and system information.
- 1How does Perplexity work?
Perplexity · Updated 2026
- 2What's New in Advanced Deep Research
Perplexity · Jul 16, 2026
- 3Creating assets with Perplexity
Perplexity · Jul 16, 2026
Last editorial verification: 2026-08-24. Prices, access limits, preview status and model routing can change after this date.