generated: '2026-08-12'
method: searched
source: >-
  https://docs.getbluma.com/guides/test-vs-production , https://docs.getbluma.com/authentication ,
  https://docs.getbluma.com/concepts/credits , https://docs.getbluma.com/concepts/rate-limits ,
  https://docs.getbluma.com/concepts/webhooks
docs: https://docs.getbluma.com/guides/test-vs-production
note: >-
  Bluma ships a genuine test environment implemented as a key-prefix mode on the same
  base URL — no separate sandbox host, no separate account. Test keys are free and
  unlimited, exercise the identical API surface, and fire real webhooks. Every value
  below is published by Bluma; nothing has been invented.
model: key-prefix mode switching
separate_host: false
base_url: https://api.getbluma.com/api/v1
key_prefixes:
- prefix: bluma_test_
  mode: test
  cost: free
  credits_charged: false
  keys_allowed: unlimited
  available_on_plans:
  - Free
  - Starter
  - Pro
  - Enterprise
- prefix: bluma_live_
  mode: production
  cost: credits charged per video
  credits_charged: true
  available_on_plans:
  - Starter
  - Pro
  - Enterprise
  note: Production keys require a paid plan.
provisioning:
  dashboard: https://app.getbluma.com/settings?tab=api
  api:
    operation: POST /api/v1/api-keys
    auth: session token
    body:
      name: string
      environment: test | production
      rate_limit_per_hour: integer (optional)
    response_fields:
    - id
    - key
    - name
    - environment
    - created_at
    warning: The secret is displayed exactly once.
test_mode_behavior:
  api_parity: 100% identical surface and semantics
  video_quality:
    max_resolution: 720p
    note: Requests for 1080p or 4K are downgraded in test mode.
    bitrate: reduced
    fps: 24 (vs 30 in production)
  watermark: 'TEST MODE overlay burned into every render'
  processing_priority: low (slower than production)
  credits: not consumed
  rate_limits: same as the account tier
  webhooks: fully functional
  restrictions:
  - no custom branding
  - no priority support
  - no SLA
fixtures:
  test_templates:
  - id: consumerclub-discord-zoomed
    name: ConsumerClub Discord Zoomed
    category: entertainment
    credits_per_video: 5
  - id: cat-explainer-v2
    name: Cat Explainer V2
    category: educational
    credits_per_video: 5
  - id: rick-morty-explainer
    name: Rick & Morty Meme Explainer
    category: entertainment
    credits_per_video: 5
  - id: steel-griffin-explainer
    name: Steel Griffin Explainer
    category: educational
    credits_per_video: 5
  note: >-
    These are the four templates Bluma's own quickstart returns from GET /v1/templates.
    They are production templates usable in test mode, not synthetic fixtures.
triggers:
- scenario: rate limit / 429 backoff
  method: >-
    Create a test key with an artificially low rate_limit_per_hour (the docs use 10) via
    POST /api/v1/api-keys, then exceed it.
  source: https://docs.getbluma.com/concepts/rate-limits
- scenario: webhook delivery
  method: Point a webhook at webhook.site, or tunnel a local endpoint with ngrok.
  source: https://docs.getbluma.com/concepts/webhooks
- scenario: insufficient credits (402)
  method: >-
    Exhaust a Free-tier allowance with production keys; Free tier permits no overage and
    fails with 402.
  source: https://docs.getbluma.com/concepts/credits
time_simulation:
  supported: false
  note: No test clock or time-travel facility is documented.
test_credentials_published: false
test_credentials_note: >-
  No shareable test API key, magic identifier, or hosted playground token is published —
  a caller must create an account to obtain even a test key. There is also no interactive
  API console or "try it" widget on docs.getbluma.com.
promotion_checklist_published: true
promotion_checklist_url: https://docs.getbluma.com/guides/test-vs-production
gaps:
- No hosted console, playground, or interactive request builder — evaluation requires
  signup.
- No published sandbox credential, so an agent cannot exercise a single call anonymously.
- No time simulation, so scheduled/daily-delivery behavior cannot be tested end to end.