Skip to main content
The Task API combines AI inference with web search and live crawling to turn complex research tasks into repeatable workflows. Define what you need in plain language or JSON, and the Task API handles the research, synthesis, and structured output—complete with citations and confidence levels.
See Pricing for a detailed schedule of rates.

​
What you can build

The Task API is designed for maximum extensibility. Create a task spec for any research need:
  • Data enrichment: Enhance CRM records, company databases, or contact lists with web intelligence
  • Market research: Generate comprehensive reports on industries, competitors, or trends
  • Due diligence: Automate compliance checks, background research, and verification workflows
  • Content generation: Create research-backed reports, summaries, and analyses

​
Prerequisites

Generate your API key on Platform. Then, set up with the TypeScript SDK, Python SDK or with cURL: 
echo "Install curl and jq via brew, apt, or your favorite package manager"
export PARALLEL_API_KEY="PARALLEL_API_KEY"

​
Core concepts

Before diving in, understand these key concepts:

Task specs

Define your research task using input/output schemas in plain language or JSON

Processors

Choose the right processor tier based on research depth and latency requirements

Research Basis

Every output includes citations, reasoning, and confidence levels for verification

​
Input and output patterns

The Task API supports flexible input/output combinations to match your use case:

​
Question in → Answer out

The simplest pattern: ask a question, get a researched answer. 
task_run = client.task_run.create(
    input="What is the founding date of the United Nations?",
    task_spec={"output_schema": "The founding date in MM-YYYY format"},
    processor="base"
)
# Output: "10-1945"

​
Question in → Report out

Generate comprehensive markdown reports with inline citations. 
task_run = client.task_run.create(
    input="Create a market research report on the HVAC industry in the USA",
    processor="ultra",
    task_spec={"output_schema": {"type": "text"}}
)
# Output: Multi-page markdown report with citations

​
Question in → Auto-structured output

Let the processor automatically determine the best output structure. 
task_run = client.task_run.create(
    input="Research the top 5 AI infrastructure companies and their recent funding",
    processor="ultra"
)
# Output: Automatically structured JSON with company profiles, funding details, etc.

​
Structured input → Structured output

Define explicit input and output schemas for precise control over data enrichment. 
task_run = client.task_run.create(
    input={"company_name": "Stripe", "website": "stripe.com"},
    task_spec={
        "input_schema": {
            "type": "json",
            "json_schema": {
                "type": "object",
                "properties": {
                    "company_name": {"type": "string"},
                    "website": {"type": "string"}
                }
            }
        },
        "output_schema": {
            "type": "json",
            "json_schema": {
                "type": "object",
                "properties": {
                    "founding_year": {"type": "string"},
                    "employee_count": {"type": "string"},
                    "total_funding": {"type": "string"}
                }
            }
        }
    },
    processor="core"
)

​
Use cases

Enrichment

Enhance structured data with web intelligence. Start with a spreadsheet or database, add new columns with researched data.

Deep Research

Conduct open-ended research without structured input. Generate comprehensive reports on any topic.

​
Next steps

​
Rate limits

See Rate Limits for default quotas and how to request higher limits.