Skip to content
localmodel.run

Audio model · whisper

Whisper small requirements

Speech to text · 244M params · int8 / fp16 · released Sep 2022. Light enough to run on CPU, no GPU required.

MIT Commercial use OK
Peak memory (fp16 (whisper.cpp))
~0.85 GB
Runs on CPU
Yes
Parameters
244M
Type
Speech to text

Run it

Runtime tools fp16 (whisper.cpp)

Whisper small runs in whisper.cpp, faster-whisper or MacWhisper at fp16 (whisper.cpp). It runs CPU-only, and the smaller tiers are fast enough for real-time use on a laptop or phone.

whisper.cppfaster-whisperMacWhisper

Which devices can run Whisper small?

Apple Silicon Macs

RAM-only laptops

iPhone & iPad

Android

NVIDIA GPUs

AMD GPUs

FAQ

How much memory does Whisper small need?

At fp16 (whisper.cpp) it consumes ~0.85 GB. It runs on CPU, so a GPU is optional.

Can Whisper small run on a phone or CPU?

Yes for CPU. The smaller tiers are light enough for real-time use, and on-device phone runtimes are available.

Can I use Whisper small commercially?

Yes. Whisper small is licensed MIT, which permits commercial use.

Notes

The phone-friendly Whisper tier (244M). whisper.cpp runtime RAM is ~852MB; the ggml model is ~466MB. Runs comfortably CPU-only and on-device on modern phones. Good accuracy/speed balance when large-v3 is too heavy. License MIT. Sources: OpenAI Whisper repo, whisper.cpp README.

Sources

Memory is a sourced peak-usage anchor at fp16 (whisper.cpp), validated 2026-06-15. See methodology.