Parallel Task API
Product by Parallel Web Systems
Parallel Task API is an asynchronous research agent designed for deep web research and data enrichment. It handles complex objectives by performing multi-hop searches across the web, delivering analyst-grade work embedded directly into business workflows. The API is capable of generating cited reports, investment memos, and competitive landscapes that require synthesis from multiple sources. It offers a wide range of processor tiers to match the difficulty of the research task, from quick metadata extraction to exhaustive multi-hour investigations.
- Languages
- English
- Latency Range
- 10 seconds to 2 hours
- Processor Tiers
- 9 levels (Lite to Ultra8x)
- Billing Model
- Per successful run
Overview
Parallel Task API is an asynchronous research agent designed for deep web research and data enrichment. It handles complex objectives by performing multi-hop searches across the web, delivering analyst-grade work embedded directly into business workflows. The API is capable of generating cited reports, investment memos, and competitive landscapes that require synthesis from multiple sources. It offers a wide range of processor tiers to match the difficulty of the research task, from quick metadata extraction to exhaustive multi-hour investigations.
Research Capabilities
- Multi-hop Research Agent: Performs multi-hop research that can run from seconds to hours for deep insights.
- Cited Structured Output: Returns cited structured output that answers complex questions not solvable in a single search.
- Scalable Reasoning Effort: Offers a range of processors from Lite for basic metadata to Ultra8x for the most difficult research.
- Production-grade Trust: Attaches provenance and calibrated confidence scoring to every output via the Basis framework.
Enrichment & Data
- Data Enrichment: Enriches databases with live, structured profiles of companies, people, and accounts.
- Per-Element Basis: Supports dot-delimited indexes for per-element citations in list outputs.
- Third-party Data Integration: Integrates with Data Connectors to bring in third-party sources like PubMed or clinical trials.
Pricing
- Lite Processor (per 1,000 runs)
- $5
- Base Processor (per 1,000 runs)
- $10
- Core Processor (per 1,000 runs)
- $25
- Ultra Processor (per 1,000 runs)
- $300
- Ultra8x Processor (per 1,000 runs)
- $2,400
Specifications
- Latency Range
- 10 seconds to 2 hours
- Processor Tiers
- 9 levels (Lite to Ultra8x)
- Billing Model
- Per successful run
Added by AntSpace from public sources (docs.parallel.ai).