Central GPU Whisper API for a small team (Docker)
whisper_api offloads transcription to one GPU server shared by several people; the author suggests the RTX 3060 12GB as the cheapest option for high-quality models.
2-10 people11-50 peopleTranscriptionUnder $1kSensitivity: High (client/confidential)
Hardware
| Item | Approx. price (as of 2026-10-16) |
|---|---|
| Server with NVIDIA GeForce RTX 3060 (12GB) - author's suggestion | Not stated in source |
| NVIDIA GeForce RTX 3060 12GB (various) (Newegg, related current price) | $449.99 |
Source prices are as stated by the source. Retailer offers: prices checked 2026-10-16, may change. Some links may be affiliate links; we may earn a commission at no extra cost to you. This never changes what we list.
Software stack
- Docker
- NVIDIA Container Toolkit
- whisper_api
- VPN for remote access (suggested)
Models used
- OpenAI Whisper models
Good at
- Multi-user central service
- Can unload model after idle time
Weak at
- Requires Linux/Docker/CUDA comfort
Rough total cost
Not stated in source.
Getting started
- Install Docker, CUDA and NVIDIA Container Toolkit.
- docker run -d -p 3001:3001 --gpus all ghcr.io/mayniklas/whisper_api:latest
- Keep it on the LAN; use a VPN for outside access.
- Optionally set UNLOAD_MODEL_AFTER_S to free VRAM.
Source & credit
README — MayNiklas/whisper_api (GitHub)
Summarized from the source; quotes are the source author's words. Check the original before buying.