Parallel Deep Research

Name: parallel-deep-research
Rating: 92
Author: nc9

Comprehensive intelligence reports from natural language queries using Parallel AI Task API.

When to Use

•User needs thorough, multi-source research on a topic
•User asks complex questions requiring synthesis of multiple sources
•User needs market research or competitive analysis
•User wants comprehensive reports with citations
•Simple web search is insufficient for the depth needed

Requirements

Environment variable must be set:

•PARALLEL_API_KEY - Parallel AI API key

Command

bash

./scripts/parallel_research.py research "query" [options]

Options

Option	Description
`-p, --processor`	Model: `pro-fast` (default), `pro`, `ultra-fast`, `ultra`
`-t, --timeout`	Max wait in seconds (default: 600)
`-f, --format`	Output: `json` (default) or `markdown`

Processors

Processor	Speed	Quality	Cost
`pro-fast`	2-5x faster	Good	Lower
`pro`	Standard	Better	Medium
`ultra-fast`	2-5x faster	Better	Medium
`ultra`	Standard	Best	Higher

Output Format

Default JSON for LLM parsing:

json

{
  "query": "What are the latest developments in fusion energy?",
  "processor": "pro-fast",
  "run_id": "run_abc123",
  "content": {
    "summary": "...",
    "key_developments": [...],
    "companies": [...],
    "timeline": [...]
  },
  "basis": [
    {
      "field": "summary",
      "excerpts": [
        {"url": "https://...", "text": "..."}
      ]
    }
  ]
}

Examples

Basic research:

bash

./scripts/parallel_research.py research "What are the key players in the AI chip market and their competitive positions?"

Higher quality (slower):

bash

./scripts/parallel_research.py research "Comprehensive analysis of renewable energy trends 2024" -p ultra

Human-readable output:

bash

./scripts/parallel_research.py research "Electric vehicle battery technology advances" -f markdown

Extended timeout for complex research:

bash

./scripts/parallel_research.py research "Deep competitive analysis of cloud providers" -t 1800

Notes

•Research can take 1-45 minutes depending on complexity
•The pro model and ultra models are expensive and can take 10-45 minutes to complete
•Query must be under 15,000 characters
•Status updates print to stderr, results to stdout
•basis field contains citations mapping content to sources