AI Agent Index
ByHeather MacAvelia·Independently reviewed·Published Mar 18, 2026·Updated Sep 23, 2026
Independently verified against live vendor data on Sep 23, 2026.

Answer engine that searches the live web and cites every claim inline, with an autonomous Computer agent on all paid plans. Free tier, Pro from $17/mo billed annually (USD).

How we scored it

Autonomy

4/5

Integrations

5/5

Pricing clarity

4/5

Evidence

5/5

Setup

5/5

The facts

Perplexity AI is an answer engine that combines large language models with live web retrieval and cites every claim inline. Rather than returning links, it synthesizes across sources and reasons over multiple retrieval steps for harder questions. The platform includes Perplexity Computer, an autonomous agent available on all paid plans that can also be started by email and, for Pro and Max subscribers with a supported NVIDIA GPU, run locally as Portable Computer. It also offers Personal Computer for local file and application tasks, the Comet browser, and Deep Research, which produces reports, spreadsheets, dashboards and websites from a single prompt. Effort Mode, added on 21 September 2026, lets personal-account users on the web choose Light, Standard, High or Ultra effort while Perplexity picks the model and reasoning level. Best fit: researchers, analysts and consultants who want cited answers quickly and are willing to work inside Perplexity's own surfaces. Perplexity both publishes and consumes MCP. It hosts a remote Model Context Protocol server at api.perplexity.ai/mcp, with a local alternative, exposing four tools for search, answers, research and reasoning to clients including Claude Code and Cursor, and a separate Computer MCP server that lets an MCP client hand multi-step tasks to Perplexity Computer. Its Agent API can also connect out to user-supplied remote MCP servers, discover their tools when a request starts, and call them during the run. The Agent API is now the default developer API alongside Search, Embeddings and Router APIs, and Perplexity states that the older Sonar Chat Completions API will be supported until September 27, 2026. Computer runs inside Microsoft 365 across Word, Excel, PowerPoint, Outlook and Teams, and Perplexity sources premium data from partners including PitchBook, Statista and Wiley. Pricing is in USD and published on the vendor's own pages. Pro is $17 per month billed annually or $20 billed monthly, and Max is $167 per month billed annually or $200 billed monthly, and both are licensed for personal use only, which the pricing page states directly. Enterprise Pro costs $34 per seat per month billed annually and adds Computer with 500 monthly credits per seat, a guaranteed no-training commitment, SSO and premium source citations. Enterprise Max costs $271 per seat per month billed annually and adds 15,000 monthly Computer credits per seat and SCIM, audit logs and data retention settings. An organization can mix the two seat types, both can be started self-serve, and a custom-priced tier adds role-based access control, a custom enterprise domain and volume discounts through sales. Perplexity is best for researchers, analysts and consultants who want fast, cited answers from the live web during daily work, and for developers who want to feed Perplexity search into other AI tools. Teams that want a general assistant centred on writing, coding and office work rather than on search-led research should consider Claude or ChatGPT. Researchers who need answers drawn from peer-reviewed literature should consider Elicit or Consensus, and organizations that want an assistant licensed and administered through Microsoft 365 should consider Microsoft 365 Copilot. Buyers handling sensitive material should note that consumer searches are used to improve Perplexity's models by default, with an opt-out in account settings, while Enterprise Pro and Enterprise Max data is not used for training. Current state Q3 2026: Computer reached Microsoft 365 including Teams on 28 May 2026, and Personal Computer reached all Mac users on 10 May 2026. Shared workspaces, Personal Computer for Windows and a Model Council mode arrived on 4 August 2026, Computer in Email followed on 24 August, and Effort Mode, a Skills Marketplace and Portable Computer arrived on 21 September. Perplexity lists SOC 2 Type II, HIPAA, GDPR and PCI DSS for its Enterprise plans, and PCI DSS sits outside our approved certification list. It is also FedRAMP Certified at Class B (Low) through the FedRAMP 20x pathway, authorized 1 February 2026 and scoped to Perplexity Enterprise and the API Platform rather than to consumer plans. More than 50,000 organizations use Perplexity Enterprise, and named customers include Latham and Watkins, Gunderson Dettmer, Bridgewater, PayPal, AlixPartners, Montana DNRC and Woven Health Collective.

Pricing

freemium · $17/mo annual

View pricing ↗

Segment

both

Setup

easy

Verified

Sep 23, 2026

Transparency

Mostly Public

Contract

Monthly or Annual

Data training

Opt-out

Human in loop

Optional

Capabilities

deep-researchweb-searchcitationsdata-analysismultilingualautonomous

Pros & Limitations

Editorial assessment

Pros

  • ✓Perplexity both publishes MCP servers, one for its search and research tools and one that hands tasks to Computer, and consumes external MCP servers through its Agent API, so it can feed cited research into other AI tools and pull third-party tools into its own runs.
  • ✓Paid plans let users choose between frontier models including GPT-6 Sol, Gemini 3.8 Flash, Claude Sonnet 5, Kimi K3, GLM 5.3 and Grok 4.6, and Effort Mode can pick the model and reasoning level for each Computer task.
  • ✓Every answer carries inline citations, and Computer extends the same sourcing into Microsoft 365 documents and autonomous multi-step web tasks on all paid plans.

Limitations

  • ⚠Perplexity states that its original Sonar Chat Completions API is supported only until September 27, 2026, so developer integrations built on it need migrating to the Agent API.
  • ⚠Pro and Max are licensed for personal use only, so business buyers need Enterprise seats from $34 per seat per month billed annually, and SCIM, audit logs and data retention settings start at Enterprise Max or a custom-priced plan.
  • ⚠Consumer searches are used to improve Perplexity's models by default and the opt-out must be set manually in account settings, while the no-training commitment applies only to Enterprise plans.

Technical Details

Deployment
webapimobile
Model architectureMulti-model orchestration across Perplexity's own models and third-party models including GPT-6 Sol, GPT-6 Astra, Gemini 3.8 Flash, Claude Sonnet 5, Kimi K3, GLM 5.3 and Grok 4.6
Autonomous rateAgentic: Computer executes autonomous multi-step browser and Microsoft 365 tasks on all paid plans, Deep Research runs multi-step investigations end to end, and the user reviews outputs rather than approving each step
Integrations
Microsoft 365Microsoft TeamsSlackPitchBookStatistaWiley
Security
SOC 2 Type IIGDPRHIPAAFedRAMP

Similar agents

Community reviews (3)

howitworks777
★★★★★
3/19/2026
heather_macavelia
★★★★★
3/19/2026

So helpful!

heatherbell3
★★★★★
3/19/2026

Rating

4.6/ 5
★★★★★

Editorial score

Community: 5.0 · 3 reviews ↓How we score this →

Recognition

Rated 4.6 ★MCP Server VerifiedListed 2026
Embed these on your site →

Industries

LegalFinanceHealthcarePublic SectorConsulting

Built for

B2BEnterpriseSaaSStartups

Leave a review

Never displayed publicly.

Agent Stacks

See workflow stacks that feature Perplexity AI.

Compare

Related Content

Is this your tool?

Claim this listing free to get your embeddable badges, a review collection link, and the Claimed badge on your listing.

Claim listing →

Free · Every Two Weeks

AI Agent Price & Rating Tracker

Price changes, new agent launches, acquisitions, and rating updates across the AI agent market. Verified against live vendor data, not vendor marketing.

No spam. Unsubscribe anytime. We never share your email.