YOU ARE VIEWING ONE ITEM FROM THE AICRIER FEED

Perplexity Decisions API Adds Fast Web Search

AICrier tracks AI developer news across Product Hunt, GitHub, Hacker News, YouTube, X, arXiv, and more. This page keeps the article you opened front and center while giving you a path into the live feed.

// WHAT AICRIER DOES

7+

TRACKED FEEDS

24/7

SCRAPED FEED

Short summaries, external links, screenshots, relevance scoring, tags, and featured picks for AI builders.

Perplexity Decisions API Adds Fast Web Search
OPEN LINK ↗
// 1h agoPRODUCT UPDATE

Perplexity Decisions API Adds Fast Web Search

Perplexity is adding Fast Search-backed web grounding to the Decisions API, enabling batch-style classification, routing, and scoring with fresh evidence while returning probabilities instead of prose. Grounding costs $0.001 per search, while Decisions input costs $0.02 per million tokens on the 27B model or $0.01 on Flash.

// ANALYSIS

This is a smart API-composition move: developers can connect live retrieval directly to structured decision pipelines without building a separate search-and-parse loop. The caveat is that retrieval quality, query construction, and evidence costs now become part of every decision threshold.

  • –Web grounding runs one Fast Search per question and returns the retrieved evidence alongside probability-based answers.
  • –Up to 100 grounded questions can be submitted in one request, making the feature useful for high-volume classification and routing.
  • –The 27B model groups related questions and reuses state efficiently; Flash is cheaper but processes questions independently.
  • –Fast Search costs $1 per 1,000 requests versus $5 for standard Search API calls, targeting routine and latency-sensitive retrieval.
  • –Developers should reserve standard web search for difficult or ambiguous queries where retrieval depth matters more than speed.
// TAGS
decisions-apisearchapistructured-outputautomationinference

DISCOVERED

1h ago

2026-11-04

PUBLISHED

1h ago

2026-11-04

RELEVANCE

9/ 10

AUTHOR

AravSrinivas