Sonar
Neural Network
Sonar is a fast, lightweight Perplexity model with citations and source customization: Sonar API for Q&A scenarios.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Sonar?
Sonar is a lightweight Perplexity model built for speed and easy integration: it answers questions and provides citations with sources, and the source set can be customized for your topic. The vendor describes it as an option for companies needing simple Q&A functionality without heavy overhead. In BotHub, Sonar is available without a VPN or foreign card; payment is made with Russian cards in rubles on a pay-as-you-go basis: only consumed tokens are charged, and they do not expire. With over 250 models in one window, comparing answers is easy, and the unified OpenAI-compatible API allows switching models in a multi-model system without rewriting integration; chats are encrypted via AES-GCM, and businesses have access to contracts, invoices, EDI, and an admin panel with limits. Connect Sonar to an in-product help assistant, gather facts with links for analysts and editors, answer common customer questions, verify individual claims before publication, and add a Q&A block to your site.Frequently asked questions about Sonar
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.
Sonar by Perplexity is a text model with native web search: it accesses the web and builds answers based on findings. Suitable for market research, fact-checking, preparing reports, and digests. Accepts images and documents as input; context window is 127,072 tokens.
Up to 114,364 tokens in a single response — enough for a comprehensive analytical report, detailed topic breakdown, or documentation draft. The practical volume depends on the task and prompt: the model won't stretch a short report, but will provide long material without cutting off mid-sentence.
A separate reasoning mode is not noted in our data, so we won't promise it. However, Sonar relies on web search: it first gathers sources, then formulates an answer. If you need visible logic, ask the model in your prompt to break down conclusions step-by-step and cite its sources.
Function calling and strict JSON mode are not noted in our data for Sonar, so please test behavior for your production scenario. You can request a structured response in your prompt, but it's safer to handle schema validation on your end. The API in BotHub is OpenAI-compatible.
You can upload images and documents: the model will analyze screenshots, diagrams, tables, or files and answer based on the content. Audio and video input formats are not supported. If you need to work with audio or video, switch to a specialized model in the same BotHub interface.
We have no data on specific language support for this model, so we won't provide ratings — it's faster to test with your real query. Access to Sonar via BotHub works from Russia without a VPN or foreign card, payment is made with Russian cards, and you can switch models in a couple of clicks.