generated: '2026-07-19'
method: searched
source: https://docs.inceptionlabs.ai/get-started/get-started
derived_from: openapi/inception-openapi-original.json
authentication:
  style: bearer-token
  header: 'Authorization: Bearer <api_key>'
  env_var: INCEPTION_API_KEY
  docs: https://docs.inceptionlabs.ai/get-started/authentication
  see: authentication/inception-authentication.yml
compatibility:
  openai_compatible: true
  base_url: https://api.inceptionlabs.ai/v1
  note: >-
    Drop-in replacement for OpenAI chat/completions request and response shapes;
    existing OpenAI SDKs work by changing base URL, model string, and key.
idempotency:
  supported: false
  note: >-
    No Idempotency-Key header or idempotent-retry contract is documented. Chat,
    FIM, and edit completions are inference calls (non-persistent); safe to retry
    on 429/503 with exponential backoff.
pagination:
  supported: false
  note: List endpoints (models) return the full set with no pagination parameters.
streaming:
  supported: true
  transport: server-sent-events
  trigger: 'set stream=true'
  chunk_objects:
  - ChatCompletionChunk
  - TextCompletionChunk
  diffusion_mode:
    param: diffusing
    note: >-
      Diffusion mode streams intermediate denoising states so a client can
      visualize how text is iteratively refined into the final answer.
  docs: https://docs.inceptionlabs.ai/capabilities/streaming
structured_output:
  supported: true
  mechanism: response_format (json_object / json_schema)
  docs: https://docs.inceptionlabs.ai/capabilities/structured-outputs
tool_calling:
  supported: true
  scope: chat/completions only (not FIM or edit)
  docs: https://docs.inceptionlabs.ai/capabilities/tool-use
reasoning_controls:
  params:
  - reasoning_effort
  - reasoning_summary
  - reasoning_summary_wait
  instant_mode: 'reasoning_effort=instant for near-realtime, low-latency turns'
  docs: https://docs.inceptionlabs.ai/capabilities/instant
request_tracing:
  request_id_header: null
  note: No documented request-id / trace header.
error_envelope:
  shape: '{ "error": { message, type, param, code } }'
  format: openai-error-envelope (not RFC 9457)
  see: errors/inception-problem-types.yml
rate_limit_signaling:
  headers: []
  note: 429 rate_limit_error signals throttling; no RateLimit headers.
  see: rate-limits/inception-rate-limits.yml