generated: '2026-07-19'
method: searched
source: https://interfaze.ai/docs
authentication:
  style: bearer-api-key
  header: 'Authorization: Bearer <INTERFAZE_API_KEY>'
  key_management: https://interfaze.ai/dashboard
  ref: authentication/interfaze-authentication.yml
idempotency:
  supported: false
  note: >-
    No idempotency-key header/parameter is documented. /chat/completions is a stateless generation
    call; retries are governed by the retry-safe 5xx/429 error classes rather than an idempotency key.
pagination:
  supported: false
  note: Single generation endpoint; no list operations, so no pagination surface.
streaming:
  supported: true
  style: server-sent-events
  param: 'stream: true'
  terminator: 'data: [DONE]'
  reasoning_stream: Reasoning tokens stream first inside <think>...</think> when reasoning_effort is set.
structured_output:
  supported: true
  mechanism: response_format json_schema (strict) following the OpenAI structured-output spec.
tool_calling:
  supported: true
  mechanism: OpenAI-compatible tools[] / tool_choice function calling.
precontext:
  header: 'x-show-additional-info: true'
  note: >-
    Verifiable raw outputs from internal tasks (ocr, stt, translate, web_search, scraper,
    object_detection, guardrails, code_sandbox) are returned in a response `precontext` array; set
    the header to inline precontext at the start of a streamed response.
versioning:
  style: uri-path
  current: v1
  ref: lifecycle/interfaze-lifecycle.yml
error_envelope:
  shape: '{ "error": { "message": string, "type": string, "code": string } }'
  format: openai-error-envelope
  ref: errors/interfaze-problem-types.yml
rate_limiting:
  standard_limit: 50 requests/second
  higher: Available on request / enterprise (unlimited).
  signal: 429 rate_limit_error; retry with exponential backoff.
  ref: https://interfaze.ai/docs/limits
size_limits:
  max_request_duration: 5 minutes
  direct_file_upload: 20 MB
  url_referenced_file: 80 MB
  context_window: 1,000,000 tokens
  max_output: 32,000 tokens