generated: '2026-07-20' method: searched source: https://mirelo.ai/api-docs docs: https://mirelo.ai/api-docs api_base: https://api.mirelo.ai/v2/ conventions: authentication: style: bearer-token header: Authorization format: 'Bearer sk-...' ref: authentication/mirelo-authentication.yml versioning: style: uri-path-and-model api_version: v2 model_version_in_path: true example: /v2/text-to-sfx/v1.6/sync notes: >- The HTTP API is versioned in the path (v2). The generation model version is ALSO pinned in the path segment (e.g. v1.6), so callers select an exact model revision per request. execution_modes: - mode: sync description: >- Synchronous generation endpoints (.../sync) return the finished result inline. Long generations may time out around ~60s; use the async mode for longer jobs. - mode: async description: >- Asynchronous job submission returns a job the caller polls for completion; used for longer/looping generations. preflight: description: >- GET .../preflight returns a cost estimate (credits) and an estimated processing time (estimated_ms) before committing a billable generation. response_fields: [credits, estimated_ms] billing: unit: credits signaling: >- Generations debit account credits; SFX 1.6 costs ~10 credits/sec and Audio-to-MIDI ~2.5 credits/sec. Preflight returns the credit cost up front. request_format: content_type: application/json response_format: content_type: application/json result_field: result_urls notes: Successful generations return audio asset URLs (e.g. cdn.mirelo.ai/...). idempotency: supported: false notes: >- No idempotency-key header or parameter is documented as of this capture. (No `type: Idempotency` pointer is emitted.) rate_limits: documented: false notes: No explicit rate-limit headers or quotas are published; usage is governed by plan credit balances. pagination: documented: false cross_links: authentication: authentication/mirelo-authentication.yml lifecycle: lifecycle/mirelo-lifecycle.yml changelog: changelog/mirelo-changelog.yml