Fact · Speech recognition
OpenAI's README says its English-only .en models tend to perform better for English applications, especially the tiny and base sizes.
Checked against its source on October 17, 2026
In the source's own words
“The `.en` models for English-only applications tend to perform better, especially for the `tiny.en` and `base.en` models.”
openai/whisper: README (raw text), OpenAI, GitHub (openai/whisper), 2026, Available models and languages.
Open the original- What it measures
- The README's model table lists an English-only version of the tiny, base, small and medium models.
- What it does not show
- The README says the difference becomes less significant for small.en and medium.en. It does not give error rates for these models in the README itself.