Perplexity
Perplexity AI · Research · General AI
An answer engine that searches and cites as it responds. Profile settings personalize responses; Spaces group research with shared instructions; paid tiers choose among frontier models.
Best for: Cited, source-first research answers; your profile shapes how findings are presented.
| Models | Multiple frontier models (selectable on paid tiers) |
|---|---|
| Free plan | Yes |
| Paid plans | Pro tier adds model choice and higher limits |
| API | Yes |
| Custom instructions | Yes, profile personalization prompt; per-Space instructions |
| Memory | Profile-based personalization |
| File upload | Yes |
| Web search | Always. That's the product |
| Context window | Model-dependent |
| Apps | Web · iOS · Android · desktop |
Facts last reviewed 2026-08-10, capabilities and plans change frequently; verify on the official site before relying on a detail.
Models, limits and pricing
| Current models | Sonar and Sonar Pro, Perplexity in-house search-grounded models Sonar Reasoning Pro and Sonar Deep Research, reasoning and report models Selectable frontier models: GPT-5.x, Claude 4.6-class, Gemini 3.1 Pro Model Council, runs one query across several frontier models at once |
|---|---|
| Context window | Not published |
| Parameters | Not disclosed most closed frontier models do not publish parameter counts; we do not repeat rumored figures. |
| Modalities | Text, image and file in; text out; image generation; Comet browser agent |
| Consumer pricing |
|
| API pricing | Sonar: $1 / 1M input, $1 / 1M output; Sonar Pro: $3 / 1M in, $15 / 1M out |
Pricing and model line-ups verified 2026-08-11 from source 1 ↗ · source 2 ↗ · source 3 ↗. These change often. The vendor page is authoritative.
Notable
- Every answer carries citations
- Spaces: research workspaces with instructions