specification: API Commons Rate Limits specificationVersion: '0.1' schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits provider: Unisound providerId: unisound generated: '2026-07-21' method: searched created: '2026-07-21' modified: '2026-07-21' tags: - Rate Limiting - Artificial Intelligence description: Unisound Token Hub enforces per-model RPM (requests per minute) and TPM (tokens per minute, input + output combined) limits; the API refuses further requests when a threshold is exceeded within the window. No rate-limit response headers are documented. On the legacy AI Open Platform, error code 20206 signals concurrency limit exceeded and 20207 quota exhausted. sources: - https://maas.unisound.com/docs/api/rate-limits limits: - name: Text models (u2, u2-med, kimi-k3, glm-5.2) requests scope: model metric: requests_per_minute limit: 100 timeFrame: minute - name: Text models (u2, u2-med, kimi-k3, glm-5.2) tokens scope: model metric: tokens_per_minute limit: 10000000 timeFrame: minute - name: Async speech transcription tasks (v1_audio_asr_tasks, u2-asr) scope: interface metric: requests_per_minute limit: 20 timeFrame: minute - name: Async speech synthesis tasks (v1_audio_speech_tasks, u2-tts / u2-tts-clone) scope: interface metric: requests_per_minute limit: 20 timeFrame: minute - name: Voice clone (v1_audio_voices_clone, u2-tts-clone) scope: interface metric: requests_per_minute limit: 20 timeFrame: minute - name: Document parsing tasks (v1_files_parser_tasks, u1-ocr) requests scope: interface metric: requests_per_minute limit: 20 timeFrame: minute - name: Document parsing tasks (v1_files_parser_tasks, u1-ocr) tokens scope: interface metric: tokens_per_minute limit: 100000 timeFrame: minute - name: OCR image extraction (v1_ocr_image_extract, u1-ocr / u1-ocr-med) requests scope: interface metric: requests_per_minute limit: 20 timeFrame: minute - name: OCR image extraction (v1_ocr_image_extract, u1-ocr / u1-ocr-med) tokens scope: interface metric: tokens_per_minute limit: 100000 timeFrame: minute