name: Unify Rate Limits
description: >
  Unify applies rate limits at the API gateway level to ensure fair usage across
  all consumers and protect the reliability of provider routing. Because Unify
  aggregates across 100+ upstream LLM providers, effective throughput depends on
  both Unify's own gateway limits and the underlying limits of the target model
  and provider. Unify's routing layer manages fallbacks and load balancing to
  maximize throughput within these constraints. Specific numeric limits are
  credit-tier dependent and available via the console. The platform itself helps
  users avoid hitting individual provider rate limits through intelligent routing
  and load distribution.
specificationVersion: '0.1'
limits:
  - name: API Request Rate
    description: >
      Requests per minute to the Unify gateway API. Limits scale with account
      tier and credit balance. New accounts with free credits operate at a
      standard rate suitable for development and testing. Higher throughput is
      available for accounts with active credit balance.
    scope: account
    window: 60 seconds
    unit: requests
    tier: standard
    notes: >
      Exact RPM limits are tier-dependent and visible in the console dashboard.
      Contact Unify for high-throughput production limits.
  - name: Concurrent Requests
    description: >
      Maximum number of simultaneous in-flight requests to the Unify gateway.
      The platform supports parallel routing to multiple providers simultaneously
      for comparison and fallback scenarios.
    scope: account
    unit: concurrent requests
    tier: standard
    notes: >
      Concurrency limits scale with credit tier. The routing layer may fan out
      a single request to multiple providers for benchmarking or fallback.
  - name: Provider-Level Limits
    description: >
      Each upstream provider (OpenAI, Anthropic, Mistral, etc.) enforces its
      own rate limits based on the credentials used. Unify uses shared provider
      API keys for free/standard tier routing. Users can supply their own
      provider API keys in the console to use their own provider rate limit
      quotas, bypassing shared pool constraints.
    scope: provider
    unit: varies by provider
    notes: >
      Bringing your own provider API keys (BYOK) is supported for all major
      providers and recommended for production workloads requiring high throughput.
  - name: Message Attachment Size
    description: >
      File attachments sent via the API messaging endpoint are limited to 10 files
      per message. This applies to the Unify assistant messaging REST API endpoint.
    scope: request
    limit: 10
    unit: files per message
  - name: Credit Balance Throttling
    description: >
      API access is gated on account credit balance. When credits are exhausted,
      requests return a 402 Payment Required response until the account is topped
      up. There is no hard rate limit based on time window alone — sustained access
      requires maintaining a positive credit balance.
    scope: account
    unit: USD credits
    notes: >
      Monitor credit usage via the console dashboard at console.unify.ai.
urls:
  - type: Documentation
    url: https://docs.unify.ai
  - type: Console
    url: https://console.unify.ai