generated: '2026-08-14'
method: searched
source: https://docs.parallel.ai/getting-started/pricing
docs:
- https://docs.parallel.ai/getting-started/pricing
- https://parallel.ai/pricing
model: pay-as-you-go
currency: USD
plan_count: 1
plans:
- name: Pay as you go
  price: 0
  currency: USD
  billing_period: usage
  self_serve: true
  signup: https://platform.parallel.ai
  included_quota: >-
    Up to 5,000 requests per month free. Parallel also advertises up to $80 in signup
    credit plus $5 in free credits per month.
  overage: Metered per request/run at the published per-API rates below.
  limits_ref: rate-limits/parallel-web-systems-rate-limits.yml
  note: >-
    Parallel publishes no named tier ladder (no Free/Pro/Enterprise). There is a single
    self-serve pay-as-you-go account with per-API metered pricing; volume/limit increases
    are handled by contacting support@parallel.ai.
enterprise_contact: support@parallel.ai
metered_pricing:
- api: Search
  unit: per 1,000 requests (default 10 results)
  rates:
  - mode: turbo
    usd_per_1000: 1
  - mode: basic
    usd_per_1000: 5
  - mode: advanced
    usd_per_1000: 5
  additional: $1 per 1,000 additional page results & excerpts
- api: Extract
  unit: per 1,000 URLs
  usd_per_1000: 1
- api: Task
  unit: per 1,000 Task Runs, by processor (fast variants priced the same as standard)
  rates:
  - processor: lite
    usd_per_1000: 5
  - processor: base
    usd_per_1000: 10
  - processor: core
    usd_per_1000: 25
  - processor: core2x
    usd_per_1000: 50
  - processor: pro
    usd_per_1000: 100
  - processor: ultra
    usd_per_1000: 300
  - processor: ultra2x
    usd_per_1000: 600
  - processor: ultra4x
    usd_per_1000: 1200
  - processor: ultra8x
    usd_per_1000: 2400
  note: Billed per Task Run (row), not per output field. Failed runs are not billed.
- api: Responses
  unit: per 1,000 requests, by reasoning.effort
  rates:
  - effort: low
    usd_per_1000: 10
  - effort: medium
    usd_per_1000: 50
  - effort: high
    usd_per_1000: 250
  note: Only successful responses are charged.
- api: Monitor
  unit: per 1,000 executions (CPM)
  rates:
  - processor: lite
    usd_per_1000: 3
  - processor: base
    usd_per_1000: 10
- api: FindAll
  unit: fixed cost per run plus per-match cost, by generator
  rates:
  - generator: preview
    fixed_usd: 0.10
    per_match_usd: 0.00
  - generator: base
    fixed_usd: 0.25
    per_match_usd: 0.03
  - generator: core
    fixed_usd: 2.00
    per_match_usd: 0.15
  - generator: pro
    fixed_usd: 10.00
    per_match_usd: 1.00
  note: Enrichments add their own per-match cost based on the Task API processor selected.
- api: Entity Search
  unit: per 1,000 requests (default 100 results)
  usd_per_1000: 5
  additional: $0.05 per 1,000 additional results
- api: Chat
  unit: per 1,000 requests, by model
  rates:
  - model: speed
    usd_per_1000: 5
  - model: lite
    usd_per_1000: 5
  - model: base
    usd_per_1000: 10
  - model: core
    usd_per_1000: 25
  note: Research models (lite/base/core) wrap Task API processors and share their pricing.
free_surfaces:
- name: Search MCP
  url: https://search.parallel.ai/mcp
  detail: >-
    The hosted Search MCP server is free to use anonymously with no API key at lower rate
    limits; a Parallel API key passed as a Bearer token unlocks higher limits.
- name: Playgrounds
  detail: >-
    Hosted try-it consoles at platform.parallel.ai/play/{task,search,extract,monitor,find-all}.
notes: >-
  Read verbatim from the published pricing reference. Parallel is purely usage-priced -
  plan_count is 1 because there is exactly one self-serve commercial arrangement
  (pay-as-you-go), not because tiers were not found. All figures are the provider's own
  published rates at the date above.