Sonar Search

Sonar is Perplexity's family of search-enabled AI models designed to generate responses using information retrieved from the web. It is useful when an AI model needs current information rather than relying only on the knowledge contained within the model itself.

In Fetch Hive, Sonar Search enables web search capabilities for Perplexity Sonar models used through the API. This allows a Sonar model to retrieve current information while generating its response, making it useful for research, current-event analysis, fact gathering, and other tasks that depend on fresh web data.

Use Cases

Current information research

Use a Perplexity Sonar model with search enabled to answer questions that depend on recent or changing information, such as market developments, company updates, product releases, or industry news.

Search-grounded AI responses

Give Sonar models access to web search so generated responses can incorporate information found online instead of relying entirely on existing model knowledge.

Research workflows

Use Sonar Search inside multi-step workflows where a model needs to investigate a topic before its output is passed into later steps for summarisation, comparison, classification, or reporting.

Why use Sonar Search with Fetch Hive?

Sonar Search lets Perplexity models retrieve information from the web while generating a response. This makes it easier to build workflows that require current information while still using the reasoning and generation capabilities of an LLM.

Fetch Hive adds observability around those calls. You can inspect trace logs, see how Sonar Search was used within a workflow, understand what happened during each step, identify failures, and track the cost of the complete run.

FAQ

What is Sonar Search in Fetch Hive?

Sonar Search enables web search capabilities for supported Perplexity Sonar models when they are used through the Perplexity API.

Is Sonar Search a separate search engine?

Not in this context. Sonar Search refers to enabling search capabilities directly within Perplexity's Sonar models so they can retrieve web information while generating a response.

When should I use Sonar Search?

It is useful when a Sonar model needs current or web-based information, including research, market analysis, current events, company research, and fact gathering.

Can Sonar Search be used inside Fetch Hive workflows?

Yes. Sonar models with search enabled can be used as part of larger Fetch Hive workflows alongside other models, tools, agents, and integrations.

Can Fetch Hive track the cost of Sonar Search calls?

Yes. Fetch Hive tracks workflow costs so you can understand the cost of model calls and other operations involved in each run.