Vendor Profile · Illustrative prototype
SerpAPI
SerpAPI 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
SerpAPI ranks #6 of 6 for agent web-search accuracy at 73.1% hit@5, and is not yet agent-ready in live onboarding trials.
hit@5
73.1%
web search
cost / correct
$0.0102
web search
agent-ready
No
onboarding harness
§ A
Web Search — SerpAPI
| Metric | SerpAPI | Leaderboard best | Direction |
|---|---|---|---|
| hit@1 % | 57.0 | Exa 71.2 | higher is better |
| hit@5 % | 73.1 | Exa 87.6 | higher is better |
| fresh<30d % | 66.0 | Tavily 86.4 | higher is better |
| retrievability h | — | Tavily 7.8 | lower is better |
| cost/correct $ | $0.0102 | Serper $0.0041 | lower is better |
| p50 latency ms | 540 | Brave 301 | lower is better |
§ C
SerpAPI vs Exa SerpAPI vs Tavily SerpAPI vs Serper SerpAPI vs Brave SerpAPI vs FirecrawlCompare 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.