Low, medium, and high presets use Fast Search
By Steven Van ·
Search calls on the affected presets now cost $1 per 1,000 invocations, plus model tokens, and run about 800 ms faster.
Web searches under the low, medium, and high Agent API presets in Perplexity now cost $1.00 per 1,000 invocations, down from $2.50, plus model tokens. Perplexity says search is about 800 ms faster.
What changes in the presets
These three presets now set search_type: "fast" for web_search, matching the fast preset. The xhigh preset continues to use standard web search. Models, prompts, tool budgets, and step limits are unchanged.
Choosing the search type
Developers can keep standard web search under any preset by setting search_type: "web" on the web_search tool in their request. If search_type is omitted, the tool inherits the preset’s search type when configured; otherwise, it uses "web".
The Web Search documentation recommends Fast Search for day-to-day agent loops and high-volume tool calls, and standard web search for rare, difficult, or ambiguous questions.