Sonar Pro Search
Neural Network
Sonar Pro Search is Perplexity's agentic search: Pro Search mode for deep reasoning and detailed analysis.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Sonar Pro Search?
Sonar Pro Search is a Perplexity model with the agentic Pro Search mode. It doesn't just provide an answer; it sequentially searches, compares findings, and builds reasoning, making it suitable for tasks requiring step-by-step topic analysis rather than brief summaries. In BotHub, the model is available without a VPN or foreign card, and you pay in rubles as you go—for consumed tokens, with no subscription. Over 250 models are available in the same window for easy switching, and a unified OpenAI-compatible API allows you to integrate search into your product and switch models later without rewriting the integration. Data is encrypted, and for companies, we offer contracts, invoices, electronic document management, and an admin panel with limits. Sonar Pro is useful for analysts to break down markets and competitors, for marketers to create reviews and content plans based on gathered data, for researchers to explore topics, for product teams to quickly gather inputs before making decisions, and for developers to embed agentic search into their own assistants.Frequently asked questions about Sonar Pro Search
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
This is a Perplexity text model focused on search scenarios: it answers questions based on found sources, analyzes uploaded documents and images, and reasons before answering. A 200,000-token context allows it to be used for research, market reviews, fact-checking, and analyzing long reports.
In one response, the model generates up to 8,000 tokens, which is several pages of text: a detailed analytical breakdown, an article, or a large code snippet. If you need more volume, ask for a continuation in the next message: the 200,000-token context allows it to keep the entire previous text in the dialogue.
Yes, this is confirmed by our data: the model reasons before answering and breaks the task down into steps. The effect is noticeable in comparisons, calculations, and questions where information from different sources needs to be synthesized. The response comes slightly slower, but the logic of the conclusions is easier to trace and verify.
Our data does not explicitly mark this, but that does not mean the capability is absent. The model is available via the unified OpenAI-compatible BotHub API, so it is easiest to test the behavior with your own schemas. You can describe the required format directly in the prompt and validate the result on your end.
You can send images and documents: the model will read a screenshot, diagram, or file and answer with text. Audio and video are not marked in our data, so for audio or video file analysis, choose a model with suitable capabilities in BotHub; switching takes just a couple of clicks.
The language of prompts and responses is not specified in our data, so we make no exact promises. The practice is simple: send your typical request and evaluate the result. Other models are available nearby in BotHub, so you can compare answers for the same task in one window, without a VPN or foreign card.