generated: '2026-09-06'
method: searched
source: https://www.edenai.co/docs/v3/general/sandbox
docs:
  - https://www.edenai.co/docs/v3/general/sandbox
  - https://www.edenai.co/docs/v3/general/custom-api-keys
name: Eden AI Sandbox
description: >-
  Eden AI implements test mode as a TOKEN TYPE rather than a separate host or a set of magic test
  values. A sandbox token accepts the identical request shape on the identical endpoints and returns
  simulated responses that match the real response structure — no provider call is made and no
  credits are consumed. Swapping the token is the only change between development and production.
mode_switch:
  mechanism: token-type
  same_base_url: true
  base_urls: [https://api.edenai.run, https://api.eu.edenai.run]
  note: There is no separate sandbox host, no test-mode header, and no key prefix that distinguishes the two.
token_types:
  - type: api_token
    purpose: Production use
    real_provider_calls: true
    cost: pay-per-use
  - type: sandbox_api_token
    purpose: Testing and development
    real_provider_calls: false
    cost: free
    returns: Mock responses matching the real response structure
creation:
  dashboard: 'https://app.edenai.run/ — API Keys'
  api:
    method: POST
    url: https://api.edenai.run/v2/user/custom_token/
    operationId: user_root_create
    body: {name: dev-testing, token_type: sandbox_api_token}
    spec: openapi/_original/eden-ai-user-management-openapi.json
usage:
  header: Authorization
  format: 'Bearer <sandbox token>'
  note: Identical to production usage.
test_values:
  published: false
  note: >-
    Eden AI publishes no test cards, test bank accounts, magic identifiers or fixture triggers —
    the mock is generated server-side from the request, so there are no published test values to
    record. Nothing has been invented to fill this section.
recommended_for:
  - CI pipelines
  - Local development against the real request/response shape
  - Validating integration wiring before spending credits
console:
  url: https://app.edenai.run/
  note: Dashboard with model catalog, monitoring, key management and guardrails.