Build fluency in the language of Perplexity's search-augmented Sonar models.
0 / 5 completed
1 / 5
At standup, a dev wants API answers grounded in live web results with citations. Which Perplexity model family fits?
The Sonar model family is Perplexity's search-augmented models that fetch live web content and return grounded, cited answers. This differs from a standard LLM answering from parameters alone. It is the core offering of the Perplexity API.
2 / 5
During a design review, the team needs the response to list the web pages used to justify each claim. Which field provides this?
Sonar responses include a citations field listing the source URLs backing the answer. This lets an app display references for transparency and verification. It is a defining feature versus non-grounded chat completions.
3 / 5
In a code review, a dev wants to restrict search results to recent pages only. Which parameter controls this?
Sonar exposes a recency filter (such as searching only the last day, week, or month) to bias results toward fresh content. This is useful for news or fast-moving topics. Without it, results may include stale pages.
4 / 5
An incident report shows a request scoped only to a set of trusted domains. Which Sonar feature does this?
Sonar supports domain filtering, letting you allow-list or block-list specific domains the search step may draw from. This is useful for compliance or quality control on sources. It constrains grounding to trusted sites.
5 / 5
During a PR review, a teammate compares Sonar Pro to a plain LLM call. What is the main tradeoff?
Choosing Sonar trades some latency and cost for live web retrieval and verifiable citations, which a plain parametric-only call lacks. For time-sensitive or fact-checked answers this tradeoff is worthwhile. For purely creative or reasoning tasks a non-grounded model may suffice.