Since its inception in July 2022, Perplexity AI has been separated from other conversation chat boats such as Chatgpt, Gemini and Copilot, which focuses online. Where traditional AI models rely on already trained data, the perplexity to achieve the latest insights is in real time. Its new Sonar API is based on the same model as it allows real -time connection to the Internet. In principle, this method helps to provide reliable and latest sources -based answers, which increases the accuracy and authority of the results.
Two separate versions
Sonar API comes in two separate versions. First, basic, fast and economical. Therefore, it aims for users with simple needs. The purpose of the Sonar Pro version is to be more powerful. The company says it specializes in dealing with complex questions with the ability to perform multiple searches simultaneously and provide twice as much references than the basic version.
At the time of launch, one of the most interesting cases of Sonar’s use is the zoom with the zoom. The famous video conferencing platform uses this API to strengthen an AI assistant who is capable of providing direct real -time responses at meetings. With this feature, users can search the Internet with quotes without leaving their chat screen.
Aggressive positioning
Perpelus Sonar holds the AI API as the most cheapest solution in the market. The basic version is offered at a fixed rate of $ 5 per 1,000 searches, which costs an additional $ 1 dollar to be entered or produced every 750,000 words (about 1 million tokens). On the contrary, Sonar Pro, which offers more sophisticated features, charges $ 15 for each equal piece of words created while maintaining uniform cost for search.
To justify this significantly high price, Perplexity claims that Sonar Pro is left behind in terms of facts accuracy, including models offered by Google, Openai and Anthropic. These performances were measured using the Simpleqa Benchmark, which specializes in reviewing the response created by AI chat boats.
With the onset of Sonar, the Perlause can eliminate a strong uprising, because it seems that the API has everything that requires real -time AI research standards. In fact, there are currently very few AI apis that offers things like Sonar and Sonar Pro. In Openai, only in real time is connected to the Internet only chatbotChatgpt. The current operator certainly connects, but it is just to automate tasks on the web, imitate human behavior in the browser. For its part, Claude AI went in a completely different direction with the feature of its quotes. Only Gemini-1.5-PRO and Gemini-2.0-Flash can find in real time based on Google search results with search glanding APIS. Other rivals should react soon, but Perplexity AI is currently one step ahead of them.