whisper-base vs whisper-medium

Specs, VRAM requirements and download trends — updated 28 July 2026.

openai · A · Open weights
Params
70 M
Context
—
30d
4.6 M
apache-2.0 · commercial OK
openai · B · Open weights
Params
760 M
Context
—
30d
379 K
apache-2.0 · commercial OK

Specification comparison

Differences are highlighted; identical values are muted.

Specification whisper-base whisper-medium
Parameters 70 M 760 M
Architecture WhisperForConditionalGeneration WhisperForConditionalGeneration
Context length — —
Licence apache-2.0 apache-2.0
Commercial use Allowed Allowed
Languages 99 99
Downloads 30d 4,637,569 379,408
Downloads all time 55.3 M 20.7 M
Quantizations on the Hub SAFETENSORS SAFETENSORS
Gated No No
Common Voice 11.0 (reported) 131 53.87
LibriSpeech (clean) (reported) 5.0087691176193 2.9
LibriSpeech (other) (reported) 12.849362732121 5.9
First seen on the Hub 2022-09-26 2022-09-26

Download trend

Daily snapshots, last 90 days (13 Aug – 10 Nov)

4.6 M305 K
whisper-base whisper-medium

VRAM side by side

On RTX 4090 · 24 GB · 8K context unless noted

whisper-base · fp16 0.8 GB / 24 GB
✅ Runs comfortably
whisper-medium · fp16 4.0 GB / 24 GB
✅ Runs comfortably

Adjacent comparisons