name: Datamuse FinOps Profile
description: |
  Mapping of the Datamuse commercial surface to the FinOps Framework and the
  FOCUS cost/usage data specification. Datamuse exposes a generous free
  non-commercial tier (100k requests/day, no key) and a custom-priced commercial
  tier negotiated directly with the provider. The FinOps profile primarily
  applies to the commercial tier; the free tier is documented for completeness.
url: https://raw.githubusercontent.com/api-evangelist/datamuse/main/finops/datamuse-finops.yml
created: '2026-05-29'
modified: '2026-05-29'
specificationVersion: '0.1'
provider: datamuse
billedThrough: Datamuse direct billing

billingModel:
  type: negotiated-contract
  cycle: per-contract
  annualOption: true
  annualDiscountNotes: |
    Datamuse does not publish standard list pricing. Annual or multi-year
    commitments are negotiated directly with the provider.

chargeCategories:
  - category: Subscription
    description: |
      Recurring fee for commercial access — covers a negotiated daily / monthly
      quota, optional custom vocabularies, and SLA terms.
    skus:
      - id: plan.free-non-commercial
        name: Free Non-Commercial Tier
        unitPrice: 0
        unit: month
        notes: Prohibited for commercial / customer-facing applications.
      - id: plan.commercial
        name: Commercial Plan
        unitPrice: null
        unit: per-contract
        notes: Custom pricing — contact Datamuse via the API documentation page.

  - category: Usage
    description: |
      Per-request or per-million-request overage charges once the contracted
      quota is exceeded. Always negotiated rather than list-priced.
    skus:
      - id: overage.commercial
        name: Commercial Overage
        unitPrice: negotiated
        unit: request
        notes: Per-contract rate; pricing depends on volume, vocabulary, and SLA tier.

  - category: VocabularyHosting
    description: |
      Optional one-time or recurring fee to host a custom domain vocabulary
      (medical, legal, gaming, internal jargon) on Datamuse infrastructure.
    skus:
      - id: vocabulary.custom
        name: Custom Vocabulary Hosting
        unitPrice: negotiated
        unit: vocabulary
        notes: Per-contract — depends on vocabulary size, refresh cadence, and exclusivity.

focusMapping:
  ServiceName: Datamuse API
  ProviderName: Datamuse
  PublisherName: Datamuse
  ServiceCategory: AI and Machine Learning
  SubCategory: Natural Language Processing
  ChargeCategory:
    Subscription: Subscription
    Usage: Usage
    VocabularyHosting: Subscription
  PricingUnit: request
  PricingCurrency: USD
  BillingCurrency: USD
  ChargeFrequency:
    Subscription: Recurring
    Usage: Usage
    VocabularyHosting: Recurring

allocationTags:
  - tag: capability
    description: Capability invoked (word-discovery, rhyme-and-phonetics, reverse-dictionary, autocomplete).
  - tag: relation
    description: Specific lexical relation queried (rel_syn, rel_rhy, rel_trg, etc.).
  - tag: vocabulary
    description: Vocabulary used (en, es, or custom vocabulary id).
  - tag: app
    description: Internal application or product identifier consuming the API.
  - tag: environment
    description: Deployment environment (dev, staging, prod).

unitEconomics:
  - metric: cost-per-1000-requests
    description: |
      Negotiated per-contract. On the free non-commercial tier this is $0
      (subject to the 100,000 requests/day quota and prohibition on commercial use).
    benchmark: contact-datamuse

optimization:
  - id: cache-stable-metadata
    description: |
      Lexical metadata (definitions, pronunciations, syllable counts, parts of
      speech, frequencies) is effectively static per word. Cache md=dpsrf
      responses indefinitely to remove >80% of repeat traffic.
    expectedSavings: high
  - id: debounce-autocomplete
    description: |
      Debounce keystrokes client-side (~150ms) before calling /sug to avoid
      issuing one request per keypress.
    expectedSavings: high
  - id: scope-with-max
    description: |
      Use the `max` parameter to bound result-set size to what the UI actually
      renders (often 10-25), reducing bandwidth and downstream processing.
    expectedSavings: medium
  - id: batch-discovery-queries
    description: |
      For exploratory tools, batch related queries (synonyms + antonyms + triggers)
      during a single user interaction rather than per-keystroke.
    expectedSavings: medium
  - id: graduate-to-commercial
    description: |
      Production / customer-facing workloads should move off the free tier and
      negotiate a commercial agreement to avoid silent throttling and to obtain
      SLA guarantees.
    expectedSavings: risk-mitigation

governance:
  - control: classify-workload
    description: |
      Tag every Datamuse-consuming workload as either non-commercial (research,
      prototype, internal tool) or commercial (customer-facing, monetized).
      Commercial workloads must be on the commercial plan to avoid TOS violation.
  - control: monitor-429
    description: |
      Alert on HTTP 429 responses; sustained 429s on a non-commercial workload
      indicate either quota exhaustion or that the workload should be moved to
      commercial billing.
  - control: review-quarterly
    description: |
      Quarterly review of request volume by capability and allocation tag to
      surface opportunities for caching, deduplication, and commercial-tier
      renegotiation.