Sonar Large
Sonar Large is Perplexity's mid-tier web-grounded model, offering greater reasoning depth and more nuanced synthesis than Sonar Small while still combining live internet retrieval with a Llama 3.1 foundation. It is positioned for research-grade queries, multi-faceted questions, and use cases where answer quality and detail matter more than minimum cost.
The model handles complex multi-source synthesis — pulling from several live web results to construct coherent, attributed answers. It is a strong fit for developer tools, research assistants, and enterprise information retrieval where users expect substantive rather than surface-level responses.
Key Features
Deep multi-source web retrieval and synthesis
Stronger reasoning and detail compared to Sonar Small
Inline citations with source URLs for every claim
Handles complex, multi-part research queries
Llama 3.1 base optimized with Perplexity's retrieval pipeline
Balances answer quality with reasonable API cost
Ideal Use Cases
Research assistants requiring deep, multi-source answers
Competitive intelligence gathering with cited evidence
Technical documentation lookups drawing on live documentation sites
Enterprise knowledge retrieval with traceable sources
Drafting research summaries from real-time web content
Example Prompts for Sonar Large
Technical Specifications
| Provider | Perplexity |
| Category | Search |
| Modality | Text -> Text (web-grounded) |
| Context Window | 127,072 tokens |
Frequently Asked Questions
Try Sonar Large now
Start using Sonar Large instantly — 100 free credits, no credit card required. Access 750+ AI models through one platform.