Parallel Search tops Artificial Analysis search benchmark at $0.014 per task in fast mode
ArtificialAnlys · x · 2026-11-21
Artificial Analysis benchmarked two modes of @p0's Parallel Search API across 1700 search-focused tasks in its Stirrup agent harness:
- Fast (objective): scores 83 on the Search Index, tied for the top spot, at $0.014 per task ($0.005 search + $0.009 model inference)
- Advanced (objective): also scores 83, at $0.030 per task ($0.023 search + $0.007 model)
Both modes sit on the cost frontier of the Search Index vs. Cost per Task chart — no product scores higher for less. For comparison, OpenAI Web Search (medium) runs $0.05 and Brave Search (LLM context) $0.079 per task.
More from Infra
- Parallel Search uses ~82k tokens per task, second-lowest of 26 search tools benchmarked — ArtificialAnlys · 2026-11-21
- Parallel Search fast mode runs 15.5s per task, both modes on the quality-latency frontier — ArtificialAnlys · 2026-11-21
- Self-hosted cloud GPUs beat Higgsfield and Runway APIs on cost and flexibility — Al_Grigor · 2026-11-21
- NVIDIA GTC DC lineup: Ian Buck, Catanzaro and Lebaredian sessions on infra, agents, physical AI — nvidia · 2026-11-21
- KV cache economics: why a $200 Claude plan can hand out $14,000 of API-equivalent tokens and still profit — rachittshah · 2026-11-21
- Baseten breaks down its security threat model as agent swarms accelerate attacks — baseten · 2026-11-21