
Perplexity Search
Investigate a topic, compare source material and refine the question with follow-ups. Check the original pages for dates, quoted wording and support for important claims.

Start with a focused question, find a trail of sources and check what supports the answer. Use Perplexity alongside other AI models in one GlobalGPT workspace.
Curated by GlobalGPT Review Team

Use Perplexity to investigate a question, then open the original sources behind important claims. GlobalGPT’s search entry, Perplexity’s official app and its developer APIs are separate routes with their own tools and account rules.
Choose the verified Perplexity search entry in GlobalGPT. Ask for a focused answer, a date range and the evidence you need; open important sources before relying on the result.

Investigate a topic, compare source material and refine the question with follow-ups. Check the original pages for dates, quoted wording and support for important claims.
Choose by workload, then review official subscriptions, developer API prices and published benchmark conditions.
| Model / API | USD rate | Billing unit | Output / conditions |
|---|---|---|---|
| Search APIPerplexity Perplexity official documentation ↗Last verified: | $5.00 | 1,000 successful requests | Ranked results; no generated answer or token fee |
| Fast Search APIPerplexity Perplexity official documentation ↗Last verified: | $1.00 | 1,000 successful requests | Fast ranked results; no generated answer |
| Agent API web_searchPerplexity Perplexity official documentation ↗Last verified: | $0.0025 | Per standard tool invocation | Model token costs billed separately |
| Agent API fast web_searchPerplexity Perplexity official documentation ↗Last verified: | $0.0010 | Per fast tool invocation | Model token costs billed separately |
| Agent API fetch_urlPerplexity Perplexity official documentation ↗Last verified: | $0.0005 | Per invocation | Page retrieval; model/search costs separate |
Search API bills successful requests, with up to five queries in one POST counting as one request. Agent API has model-token charges plus tool usage. Sonar is retired; its former token prices are not current plan prices. GlobalGPT and app subscriptions are separate.
| Plan | Billing / listed price | Plan scope | Product |
|---|---|---|---|
| Standard (Free)Perplexity Perplexity official documentation ↗Last verified: | $0 | Basic searches and limited advanced usage | Official Perplexity app |
| ProPerplexity Perplexity official documentation ↗Last verified: | Paid app subscription | Extended Pro Search, advanced models and increased file allowances | Official Perplexity app |
| MaxPerplexity Perplexity official documentation ↗Last verified: | $200 / month or $2,000 / year | Highest individual allowances and priority capabilities | Official Perplexity app; annual option on web |
| Education ProPerplexity Perplexity official documentation ↗Last verified: | $10 / month | SheerID eligibility verification required | Official education subscription |
App-plan scope and listed rates come from the official Help Center. Max annual billing is web-only; Education Pro requires eligibility verification. App allowances, model access and tools are separate from GlobalGPT and developer API billing.
| Model / configuration | BrowseComp accuracy | DSQA F1 | Evaluation scope |
|---|---|---|---|
| Agent API fastPerplexity Perplexity official documentation ↗Last verified: | 26.67% | 53.30% | Perplexity’s shared API evaluation harness |
| Agent API lowPerplexity Perplexity official documentation ↗Last verified: | 48.67% | 64.90% | Perplexity’s shared API evaluation harness |
| Agent API mediumPerplexity Perplexity official documentation ↗Last verified: | 71.67% | 85.90% | Perplexity’s shared API evaluation harness |
| Agent API highPerplexity Perplexity official documentation ↗Last verified: | 78.00% | 89.60% | Perplexity’s shared API evaluation harness |
| Agent API xhighPerplexity Perplexity official documentation ↗Last verified: | 85.33% | 93.00% | Perplexity’s shared API evaluation harness |
Perplexity reports these two benchmarks for Agent API effort levels under the same evaluation. This is developer-API evidence, not a GlobalGPT search or consumer-app test. Quality and per-request cost depend on effort; no benchmark score is inferred for the GlobalGPT entry.
| Model / product | Use it for | Access | Product scope |
|---|---|---|---|
| Perplexity on GlobalGPTPerplexity Perplexity official documentation ↗Last verified: | Search as part of a broader multi-model workflow | GlobalGPT’s Perplexity entry and account rules | GlobalGPT search workspace and its account allowances |
| Official Perplexity AppPerplexity Perplexity official documentation ↗Last verified: | Workflows that need Perplexity’s native app features | Free or paid app access according to the current plan | Official native app, including plan-specific modes and tools |
| Perplexity Developer APIsPerplexity Perplexity official documentation ↗Last verified: | Building source-grounded answers or search into an application | Developer key, usage-based billing and API-specific tools | Agent API for answers; Search API for ranked results |
Perplexity’s app, GlobalGPT search and developer APIs are distinct. Sonar Chat Completions support ended September 27, 2026; current developer integrations use Agent API. Related GlobalGPT summarizer, proofreading and PDF tools are not described as Perplexity-powered.
Start with the basics, improve your search and source checks, or explore comparisons and developer workflows. App-specific features in linked guides belong to the route each article describes.
Understand the search workflow and the difference between an app mode and a model.
Replace the template fields with your question, audience, dates and source preferences. Ask for unknowns and conflicting evidence to stay visible.
Define the decision, time range and preferred sources. Ask for supported findings rather than a padded list.
Research Prompt TemplatesKeep the criteria and product versions consistent. Ask which claims are verified and which still need checking.
Comparison Prompt TemplatesSpecify the audience, format and evidence table. Separate facts, estimates and recommendations.
Report Prompt TemplatesAsk for the exact supporting passage and date. Mark missing evidence instead of substituting a plausible source.
Source-Verification Prompt TemplatesStart with Perplexity search, then use GlobalGPT’s related tools to read documents, summarize material and review wording. The document and writing tools use their own workflows and are not presented as Perplexity-powered.
01Open the verified Perplexity workspace, investigate a focused question and inspect the sources behind important claims.
GlobalGPT
02Condense pasted material or a supported file into key points. Review names, dates, figures and qualifications against the original.
GlobalGPT
03Upload a PDF and ask questions about its contents. Check the answer against the original document and its surrounding context.
GlobalGPT
04Review grammar, spelling and clarity in your research notes or final brief. Apply suggestions that preserve the meaning; source verification is a separate step.
GlobalGPT
Official documentation explains product and developer scope. GlobalGPT guides explain workflows; the original source remains the authority for a claim.
Model choice, access routes and evidence in one place.
Start with a focused question, investigate the returned material and inspect original sources. This page links to GlobalGPT’s Perplexity workspace.
No. Pro and Max are native Perplexity app plans with their own tools and allowances. GlobalGPT has a separate account, workspace and usage rules.
They are related GlobalGPT tools. This page does not claim that Perplexity powers them; the Perplexity card is the dedicated search entry.
Perplexity ended Sonar Chat Completions support on September 27, 2026. Follow the Agent API migration guide; synchronous and streaming compatibility requests are being reformulated, while async requires background mode.
The Search API returns ranked results and bills successful requests. Generated source-grounded answers use Agent API with separate model-token and tool charges.
No. Inspect whether each original source supports the exact claim, date and scope. The published Agent API benchmark results describe that evaluation, not a guarantee for the GlobalGPT workspace.
Start in Perplexity, check the sources and move the useful findings into your next brief, draft or decision.