Perplexity AI
Available Perplexity AI Models Overview.
Perplexity AI Models
The following Perplexity AI models are available. Each model offers distinct capabilities tailored for various applications and needs.
| Models | Generate Text | Web Search | Reasoning | Research Depth | Context Window | Max Output Tokens | Availability |
|---|---|---|---|---|---|---|---|
| sonar-pro | ✅ | ✅ | ❌ | High | 200k tokens | 8k tokens | Pro |
| sonar | ✅ | ✅ | ❌ | Medium | 128k tokens | - | Free |
Please note that some features may not be available within Giselle even if they are offered in the official API.
sonar-pro
An advanced search model offering detailed responses with enhanced citation accuracy. Optimized for multi-step queries and tasks that require deep content understanding and comprehensive sourcing. Ideal for academic research, market analysis, and complex informational tasks.
sonar
A lightweight, cost-effective search model optimized for quick, accurate answers utilizing real-time web search with citations. Suitable for everyday queries, quick fact-checking, summarizations, and general-purpose use.
Model Selection Guide
Guidelines for selecting the optimal Perplexity AI model:
- For advanced multi-step research queries with citations: sonar-pro
- For everyday quick searches and cost-effective queries: sonar
Practices for Giselle
In Giselle, you can set up multiple sonar-pro nodes to approach different perspectives, creating detailed, multifaceted research outputs similar to original deep-research reports.
For detailed specifications or additional assistance, please refer to Perplexity AI Documentation.