Perplexity: Sonar

Perplexity Sonar, created on January 27, 2025, is a lightweight, fast, and affordable AI model designed for quick question-and-answer features, now with citations and customizable sources. It demonstrates competitive response times, ranking in the 48th percentile for speed, and offers competitive pricing, placing in the 42nd percentile. Notably, Sonar exhibits exceptional reliability with a perfect 100% success rate across all benchmarks, indicating consistent and usable responses. The model's key strengths lie in its General Knowledge, achieving perfect 100% accuracy, making it the most accurate model at its price point and among models of similar speed. It also performs strongly in Ethics (99.0% accuracy) and Reasoning (74.0% accuracy, 52nd percentile). However, Sonar shows significant weaknesses in Hallucinations, with only 52.0% accuracy, suggesting it frequently fails to acknowledge uncertainty. Its Instruction Following capabilities are also limited, scoring 32.0% accuracy. While its Mathematics (88.0%) and Coding (82.0%) scores are respectable, its Email Classification (92.0%) performance is below average compared to other models. Overall, Sonar is a reliable and cost-effective option for general knowledge tasks, but users should be aware of its limitations in handling uncertainty and complex instructions.