Perplexity Sonar is an LLM built on top of Llama 3.3 70B and further trained in-house by Perplexity to optimize answer quality, factuality, and readability for search-augmented tasks. It excels in delivering fast, accurate answers grounded in real-time web data with detailed citations—making it especially effective for research, fact-checking, and obtaining up-to-date information.

Some other noteworthy features of Perplexity Sonar include its ability to provide concise responses with source attribution at high speed (up to 1,200 tokens per second), seamless integration with Perplexity’s search engine for enhanced user experience, and a context window suitable for handling complex queries.

Metric	Value
Parameter Count	70 billion
Mixture of Experts	No
Context Length	128,000 tokens
Multilingual	Yes
Quantized*	No

*Quantization is specific to the inference provider and the model may be offered with different quantization levels by other providers.