The AI Bench
OpenAI · MIT

Whisper large-v3-turbo

OpenAI’s speech-to-text model for 99 languages, fast through faster-whisper. The default choice for transcription.

How much memory it needs

Memory to run each model, from its model card or the size of its files. Smaller versions trade some quality for less memory.

ModelVersionNeedsBusiness use
Whisper large-v3-turbo (via faster-whisper)FP161.6 GBAllowed

Which computers run it

The best model in this group for each machine.

ComputerBest versionFit
Laptop or PC, no graphics card, 32 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
RTX 4060 · 8 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
RTX 3060 · 12 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
RTX 5060 Ti · 16 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
RTX 4090 · 24 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
RTX 5090 · 32 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
Mac with M4 · 16 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
Mac with M4 Pro · 48 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
Mac with M5 Max · 128 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
Ryzen AI Max+ 395 · 128 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
DGX Spark · 128 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
RTX PRO 6000 · 96 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
Mac Studio with M5 Ultra · 256 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great
Mac Studio with M3 Ultra · 512 GBWhisper large-v3-turbo (via faster-whisper) FP16Runs great

What it is good and bad at

Good at

  • 99 languages
  • Runs on CPU or GPU
  • MIT licence

Not good at

  • No speaker labels on its own

Start it

  • Whisper large-v3-turbo (via faster-whisper): runs in faster-whisper (also whisper.cpp, Transformers). Model card