← Back to directory

Parallel

Where agents find answers

Visit Parallel

Parallel is web infrastructure built for AI agents rather than humans: a family of APIs (Search, Extract, Task, FindAll, Monitor, Responses/Chat) that let agents search the live web, pull clean markdown from any URL, run cited multi-hop research, discover and structure entity datasets, and get webhook alerts when tracked pages change.

Vendor's own claim
data/enrichmentorchestration/agent runtimeresearch/evidence

Core capabilities

Vendor's own claim
  • Search API: one-round-trip natural-language web search returning ranked URLs and LLM-optimized excerpts
  • Extract API: converts public URLs, including JS-heavy pages and PDFs, into clean markdown
  • Task API: multi-hop deep research and data enrichment with citations and confidence scoring, defined by a JSON output schema
  • FindAll API: turns a natural-language objective into a structured, verified dataset of matching entities
  • Entity Search: fast synchronous people/company lookup (beta)
  • Monitor API: scheduled queries with webhook alerts when tracked web content changes
  • Responses/Chat API: OpenAI-compatible chat completions grounded in live web citations
  • Data-tool integrations: native connectors for BigQuery, Snowflake, Spark, DuckDB, Polars, Supabase, Google Sheets

API & agent surface

From official docs
Authentication
API key in the x-api-key header (ApiKeyAuth per the OpenAPI spec)
Base URL
https://api.parallel.ai
Rate limits
Default limits on POST (resource-creating) requests: Search 600/min, Extract 600/min, Tasks/TaskGroups 2,000/min, Chat 300/min, Entity Search 600/min, Monitor 300/min, FindAll 300/hour. GET requests (status/result retrieval) are not rate-limited. Higher limits available on request.
Notable endpoints
  • Search - searches the web, returns ranked URLs and excerpts
  • Extract - extracts relevant content from specific web URLs
  • Create Task Run / Retrieve Task Run Result - deep research or enrichment run against a JSON output schema, with citations
  • Create FindAll Run / FindAll Run Result - discovers and structures entity datasets from a natural-language objective
  • Fast Entity Search - synchronous ranked people/company search (beta)
  • Create Monitor / List Monitor Events - scheduled web tracking with webhook events
  • Create Response (Chat) - OpenAI-compatible chat completions grounded with web citations
MCP tools exposed
web_searchweb_fetchcreateDeepResearchcreateTaskGroupgetStatusgetResultMarkdown
Machine-readable spec
OpenAPI spec retrieved

Official skills & repos

From official docs

Published by the vendor themselves — not community forks.

  • parallel-agent-skills62 stars

    Open-source Agent Skills for web search, extraction, deep research, data enrichment, and entity discovery, installable via npx skills / the parallel-cli into Cursor, Cline, GitHub Copilot, Claude Code, and 30+ other coding agents.

From the vendor's channel

Rebuilding web search for AI agents | Parallel Web SystemsParallel Web Systems

Pricing

From official docs
PlanPriceIncludedNotable
Free$0Up to 5,000 requests/month; up to $80 signup credit plus $5 monthly free creditsSearch API, Extract API, Task API access at pay-as-you-go rates
Pay-as-you-goTask API: $0.005-$2.40/request; Search API: $0.001-$0.005 per 10 results; Extract API: $0.001/request; Responses API: $0.01-$0.25/request; Monitor API: $0.003-$0.01/request; FindAll API: $0.03-$1.00/matchAll API products, Tiered processors per product (e.g. Task API: Lite $5, Base $10, Core $25, Core2x $50, Pro $100, Ultra $300, Ultra2x $600, Ultra4x $1,200, Ultra8x $2,400 per 1,000 requests)
EnterpriseContact for pricingZero Data Retention, Data Protection Agreements, Single Sign-On

Checked 2026-07-31 against their pricing page. Vendors change pricing without notice.

What is Parallel

Vendor's own claim

Parallel (Parallel Web Systems) builds a set of task-specific web APIs aimed squarely at AI agents rather than at rendering pages for humans: Search and Extract for grounding, Task for cited deep research and structured enrichment, FindAll for entity discovery, and Monitor for continuous tracking.

Ideal customer profile

Vendor's own claim

Teams building AI agents or AI-powered products that need verified, cited, real-time web data as a primary input - GTM/enrichment tooling, research and due-diligence products, and vertical AI agents in regulated or high-stakes domains (the vendor names law, finance, and healthcare). The vendor lists Notion, Dropbox, and Harvey as customers.

Who should look elsewhere

Our read

Pricing is metered per request/match and varies widely by processor tier (Task API alone ranges from $5 to $2,400 per 1,000 requests depending on tier), so teams wanting a flat, predictable bill will find costs hard to forecast without testing tiers first. This is API infrastructure, not a UI product - teams that want a point-and-click research or enrichment tool without engineering integration work are better served elsewhere. FindAll and Monitor also charge separately from Search/Extract, so a workflow that touches several of Parallel's APIs stacks multiple metered line items.

Parallel alternatives

Our read

APIs that answer a query with retrieved web content, built to be called by an agent rather than read by a person.

Category: Search and research APIs. Membership is decided by 2 stated tests, not by category tags — see them and the full comparison.

Parallel FAQ

How is Parallel priced?

Metered pay-as-you-go per API: Task API $0.005-$2.40/request across nine processor tiers (Lite through Ultra8x), Search $0.001-$0.005 per 10 results, Extract $0.001/request, Responses $0.01-$0.25/request, Monitor $0.003-$0.01/request, FindAll $0.03-$1.00/match. A free tier covers up to 5,000 requests/month.

From official docs
Is there a free tier?

Yes - up to 5,000 requests/month free, plus up to $80 signup credit and $5/month ongoing free credit.

From official docs
What does the Task API actually return?

A structured result matching a JSON schema the caller defines, with citations and confidence scoring, from a multi-hop research process; results can be retrieved via blocking call, polling, or server-sent events.

From official docs
Does Parallel publish official agent-skill repos?

Yes - parallel-agent-skills on the vendor's GitHub org, an MIT-licensed collection of Agent Skills for web search, extraction, research, and enrichment, installable into 30+ coding agents via npx skills.

From official docs
What data-warehouse integrations exist?

Native SQL-native functions/UDFs for Snowflake, BigQuery, and Apache Spark, plus integrations for DuckDB, Polars DataFrames, Supabase Edge Functions, and a Google Sheets PARALLEL_QUERY function.

From official docs

Verify before you buy

  • Exact seat/team pricing structure beyond the per-request metered rates (no separate 'plans by seat' page found - pricing is entirely usage-based plus a custom Enterprise tier)
  • GitHub star counts for repos other than parallel-agent-skills
  • LinkedIn/X follower counts

Need help setting it up?

This page tells you whether Parallel fits. Cheetah builds and runs the system it goes into — enrichment, sequencing, CRM and the joins between them.

Book a call →

The directory stays free. There is nothing gated behind this.