Perplexity adds web grounding to its Decisions API
Perplexity added a web grounding mode to its Decisions API, which runs a search for each question and passes the results to the decision model as evidence. Each completed search costs $0.001 plus input tokens.
By Cora Whitley, GTM AI Analyst, Chief of Staff
Image: Perplexity press materials, from the announcement
What we are seeing
Perplexity described web grounding for the Decisions API in a November 4 post. Setting the mode to web_grounding makes the API run a Perplexity Fast Search query per question and add titles, URLs and page text to the request as evidence. Perplexity lists use cases such as qualifying accounts on recent news, for example whether a company raised funding in the last 90 days, tracking software dependencies and checking facts before an agent acts. The company reports accuracy on 328 true or false claims rising from 56.4 to 97.3 percent with grounding, with median latency rising to 366 milliseconds. A request can include up to 100 questions.
Discuss this with peers in the ForgeX Community- Vendor
- Perplexity
- Category
- AI Models
- Announcement type
- Feature update
Report an error
Vendors and readers can flag a factual error in this story, or ask for an image or logo to be removed. Email Cora, the editor, with the link. Image and logo removals happen within one business day. For factual corrections, Cora replies with next steps and a ForgeX lead approves the fix before it is published.
