arlen/benchOpen benchmarks for agentic consumers
SNAPSHOT web_search-2026-q3
11 JUN 2026 · BERKELEY, CA
Vendor Profile · Illustrative prototype

Brave

Brave on the arlen/bench leaderboards — agent web-search accuracy, extraction fidelity, freshness, cost per verified-correct answer, and agent-readiness, scored against golden truth.

Answer firstillustrative

Brave ranks #4 of 6 for agent web-search accuracy at 77.5% hit@5; #7 of 7 for extraction fidelity at 0.63, and is not yet agent-ready in live onboarding trials.

hit@5
77.5%
web search
cost / correct
$0.0048
web search
extraction fidelity
0.63
web extraction
agent-ready
No
onboarding harness
§ A

Web Search — Brave

snapshot web_search-2026-q3
Metric BraveLeaderboard bestDirection
hit@1 %61.7Exa 71.2higher is better
hit@5 %77.5Exa 87.6higher is better
fresh<30d %72.5Tavily 86.4higher is better
retrievability h13.9Tavily 7.8lower is better
cost/correct $$0.0048Serper $0.0041lower is better
p50 latency ms301leadslower is better
§ B

Web Extraction — Brave

snapshot web_extraction-2026-q2
Metric BraveLeaderboard bestDirection
fidelity 0-10.63Firecrawl 0.91higher is better
JS gap Δ0.34Firecrawl 0.03lower is better
block rate %11.4Apify 3.3lower is better
cost/correct $$0.0036Jina $0.0009lower is better
schema validity %Firecrawl 96.2higher is better
§ C

Compare Brave

Brave vs Exa Brave vs Tavily Brave vs Serper Brave vs Firecrawl Brave vs SerpAPI

Illustrative prototype. No verified vendor run has been published yet; every figure here is a placeholder and must not be cited as a measured result. Numbers are replaced when a snapshot’s first full run lands.