Firecrawl updates /search endpoint with extraction model

// 3h agoPRODUCT UPDATE

Firecrawl updates /search endpoint with extraction model

Firecrawl upgraded its /search endpoint with a model that extracts query-specific answer excerpts instead of returning raw web pages. This update reduces token consumption tenfold while scoring 94.7% on SimpleQA.

// ANALYSIS

Direct extraction of relevant context at the search level is a major efficiency boost for agentic workflows, dramatically slashing token costs and latency without sacrificing retrieval quality.

–Uses a custom-trained model to return concise context snippets targeted to the user query.
–Reduces token consumption by 10x compared to ingesting entire web pages.
–Outperforms competing search providers on SimpleQA with a score of 94.7%.
–Deployed live by default across all standard Firecrawl /search API requests.

// TAGS

devtoolagentweb-searchllmfirecrawlrag

DISCOVERED

3h ago

2026-07-24

PUBLISHED

8h ago

2026-07-24

RELEVANCE

9/ 10

AUTHOR

[REDACTED]

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

BENCHMARK1h ago

Cole Medin releases open-source AI agent reliability benchmark

Creator Cole Medin has released an open-source benchmark repository built to evaluate AI coding agents like Kimi K3 on real-world engineering challenges rather than sanitized leaderboards. The suite includes evaluation workflows, prompt templates, and a seven-dimension scoring rubric designed to stress-test agentic reliability and expose common failure modes.

VIDEO2h ago

Poolside unveils Pool agentic coding harness

Pool is an agentic coding harness developed by Poolside designed to execute and evaluate large language models on complex software engineering challenges. Built to manage long-horizon coding tasks, Pool supports parallel execution threads, extended reasoning traces, and continuous verification loops, enabling models like Laguna S 2.1 to systematically solve, test, and refine multi-file codebases.

OPEN SOURCE3h ago

Chat2DB simplifies SQL querying with natural language

Chat2DB is an open-source, AI-driven database management platform and SQL GUI tool that integrates natural language processing into data workflows. Supporting over 40 database engines, Chat2DB enables developers and data analysts to write, optimize, and explain SQL queries using conversational prompts while offering integrated BI dashboards, ER diagrams, and offline operation.