Pricing

Simple per-unit prices, prepaid in USDC

You pay for server-counted audio seconds and tokens, nothing else. No seats, no minimum monthly spend. Long files go through the batch API, which is where the price advantage is.

Distributed processing. Jobs run on a network of vetted consumer GPUs and Macs owned by independent contributors. Anyone running a job can see its input. Do not send personal or sensitive data. Read the data policy.

The price list is temporarily unavailable. We don't show guessed numbers while a source is down.

Why it can be cheaper

Raw compute for batch Whisper on consumer GPUs costs roughly $0.002–0.026 per audio hour. Hardware that people already own, running work in batches of long files, is what keeps the price down. Salad has shown the same thing at scale: nearly 8,000 audio hours in about 10 hours for roughly $110 all-in.

What you give up

  • Privacy from the device owner. Consumer devices have no confidential-computing hardware, so inputs are visible to whoever runs the job. Send public or non-sensitive data only.
  • Real-time latency. This is a batch product. Short clips are supported, but the economics favour long files.
  • Zero data retention. We don't claim it. Retention periods are on the data policy.