AiNews.com
Posts
Perplexity Unveils Sonar: A Blazing-Fast AI Model for Search & Accuracy

Perplexity Unveils Sonar: A Blazing-Fast AI Model for Search & Accuracy

Alicia Shapiro
February 12, 2025 • Estimated Reading Time: 5 minutes

A futuristic AI interface generating a fast, glowing search result. The modern, sleek digital screen displays minimal but structured text appearing in real-time. The background features abstract data streams and digital elements, symbolizing advanced AI-driven knowledge retrieval and speed, while maintaining a visually balanced and uncluttered look.

Image Source: ChatGPT-4o

Perplexity Unveils Sonar: A Blazing-Fast AI Model for Search & Accuracy

Perplexity has introduced Sonar, its latest in-house AI model designed to optimize answer quality and user experience. Built on Llama 3.3 70B, Sonar has been further refined to enhance factual accuracy and readability, making it a powerful option for Perplexity’s search functionality.

Perplexity Pro users can access Sonar starting today, and the model is also available via the Sonar API.

Sonar’s Competitive Edge

Through extensive A/B testing, Sonar has demonstrated superior performance compared to other models in its class, such as GPT-4o mini and Claude 3.5 Haiku. It even rivals high-end models like GPT-4o and Claude 3.5 Sonnet in user satisfaction—while running at a fraction of the cost.

A bar chart showing normalized user satisfaction scores for Sonar compared to Claude 3.5 Sonnet, GPT-4o Mini, and Claude 3.5 Haiku. Sonar ranks the highest, surpassing competing models in overall user preference and engagement.

User Satisfaction. Image Source: Perplexity

One of Sonar’s standout features is its exceptional speed. Powered by Cerebras inference infrastructure, it achieves a decoding throughput of 1,200 tokens per second, making it nearly 10 times faster than models like Gemini 2.0 Flash. This enables users to receive instant answers, making Sonar ideal for both quick searches and detailed inquiries.

A gauge-style graph displaying the speed of Sonar, achieving 1,200 tokens per second, compared to Gemini 2.0 Flash (140 tokens per second) and Claude 3.5 Sonnet (75 tokens per second). The visualization highlights Sonar’s superior processing speed for generating AI-driven responses.

Speed Tokens per Second. Image Source: Perplexity

A bar chart comparing the latency (tokens per second) of different AI models. Sonar significantly outperforms competitors like Gemini 2.0 Flash, GPT-4o mini, and Claude 3.5 Sonnet, achieving nearly 10x faster processing speeds.

Sonar Model Speed Comparison. Image Source: Perplexity

What Makes Sonar Unique?

Perplexity optimized Sonar for two key areas:

Answer Factuality – Ensures responses are grounded in reliable sources and resolves conflicting or incomplete information.
Readability – Enhances clarity and formatting, making answers concise, well-structured, and easy to follow.

Additionally, Sonar excels in industry benchmarks:

IFEval – Measures how well the model follows user instructions.
MMLU – Assesses world knowledge across a wide range of topics.

By improving these aspects, Sonar surpasses its base model (Llama 3.3 70B Instruct) and outperforms competing AI models in user instruction adherence and knowledge depth.

A table comparing the performance of various AI models, including Sonar, GPT-4o, Claude 3.5 Sonnet, and others. The table displays scores for factuality, readability, IFEval (instruction following), and MMLU (world knowledge), showing that Sonar leads in multiple categories.

Sonar Model Performance Comparison. Image Source: Perplexity

Using Sonar Today

Perplexity Pro users can make Sonar their default model in settings, and businesses can access it via the Sonar API for seamless integration.

With its blazing-fast performance, high factual accuracy, and structured readability, Sonar is positioned as an ideal AI model for everyday searches and complex queries alike.

Looking Ahead

The release of Sonar highlights Perplexity’s commitment to AI-powered search innovation, balancing speed, accuracy, and user experience. As AI models continue evolving, faster and more reliable AI assistants like Sonar will shape the future of knowledge retrieval and decision-making.

With its ability to generate instant, highly factual responses, Sonar could also emerge as a serious alternative to traditional search engines like Google, offering users concise, well-structured answers without the need to sift through multiple links. If AI-driven search continues improving, it may redefine how people find and consume information online.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.