A drop-in for OpenAI's /v1/audio/transcriptions endpoint. Same response shape. Catalog list pricing well below typical cloud list rates, with concurrency that scales as the host fleet grows.
Sample request shapes for illustration only — not a live production feed, job IDs, or measured latencies.
Accuracy and latency figures below are illustrative targets for clean-speech, short-clip workloads — not measured production telemetry. Prices are catalog list rates versus a published OpenAI Whisper list reference.
Rates and model ids match the billing catalog and /pricing; catalog changes update this page automatically. Latency and WER callouts are illustrative targets, not measured fleet SLAs. OpenAI-compat accepts Whisper-family model names; Fast (Parakeet) is selected on the native jobs API.
Multilingual Whisper launch route on Apple Silicon. Use this catalog model (or Whisper-family aliases such as whisper-1 and whisper-large-v3, both of which resolve to Standard turbo-q4) on the OpenAI-compatible path. Word-level timestamps included. Illustrative accuracy/latency targets only — not production SLAs.
English Parakeet launch SKU for clean speech where cost and latency matter. Request via POST /api/v1/jobs with backend="fluid-audio-swift" and model="parakeet-tdt-0.6b-v2" (not as an OpenAI-compat model id); omitting either falls back to the Standard Whisper route. Illustrative targets only.
If your code uses the OpenAI SDK today, change the base URL and the key. Everything else — payload shape, error codes, retries — is identical.
from openai import OpenAI client = OpenAI( api_key=os.environ["OPENAI_API_KEY"], ) resp = client.audio.transcriptions.create( model="whisper-1", file=open("interview.mp3", "rb"), response_format="verbose_json", )
from openai import OpenAI client = OpenAI( api_key=os.environ["ACORN_API_KEY"], base_url="https://api.acorncompute.com/v1", ) resp = client.audio.transcriptions.create( model="whisper-1", file=open("interview.mp3", "rb"), response_format="verbose_json", )
Tell us about your transcription workload. We review pilot access requests manually and email credentials when your account is approved.