Perplexity · reasoning
Sonar Deep Research is a reasoning model associated with Perplexity. Perplexity is the creator route for Sonar Deep Research. Compare that official price with router and host pages before you lock a gateway into production, because the same model name can be billed differently once another company serves it. The recorded context window is 128,000 tokens, with max output around 32,768 tokens. Sonar Deep Research is marked for explicit reasoning, web access. Speed tier: slow. Stored use cases: General.
Sonar Deep Research is listed at $2.00 per million input tokens and $8.00 per million output tokens (USD). That quote is the recorded API rate for this page's provider route, not a subscription seat and not a blended monthly bill. A short prompt and a long agent trace will not cost the same amount even when the sticker price stays put.
No separate access-point rows are stored beyond the Perplexity route in the slug. Search the model directory or the API price board before you assume a second host.
How to use this Sonar Deep Research page: read the price as a per-million-token quote, then open the API price directory if you need other providers for the same family. Open a related model page when you are choosing between sizes or between the creator API and a router. Open the product page when the practical question is a consumer subscription (seat, credits, or included models) rather than metered API spend. Status on this record is available, with a stored release date of 2025-03-07.
No separate access-point rows are stored for Sonar Deep Research. The recorded route is Perplexity.
Sonar Deep Research via Perplexity is listed at $2.00 input and $8.00 output per million tokens. Multiply those rates by your own token volume. Deepapi does not convert them into a monthly subscription on this page.
Sonar Deep Research is filed under Perplexity. A dedicated access-point table is empty for this id, so confirm the live console on Perplexity before you call the model in production.
Sonar Deep Research is a reasoning model from Perplexity. Vision is not marked supported. Tool use is not marked supported. Reasoning is supported. Context window: 128,000 tokens. Compare sibling model pages before you pick a larger or smaller variant.