Wav2Vec 2.0 models

32 Wav2Vec 2.0 models from Meta on Hugging Face, from 315M to 963M parameters.

Fine-tunes and community models (32)

ModelParametersPublished asVRAM neededLive GPU fitEst. $/hr
wav2vec2-large-xlsr-53-basque315MF321.4 GBV100$0.110/hr
wav2vec2-large-robust-ft-libri-960h315MF321.4 GBV100$0.110/hr
wav2vec2-xls-r-300m-hebrew315MF321.4 GBV100$0.110/hr
wav2vec2-large-xlsr-53-english315MF321.4 GBV100$0.110/hr
wav2vec2-large-xls-r-300m-albanian-colab315MF321.4 GBV100$0.110/hr
nb-wav2vec2-300m-bokmaal-v2315MF321.4 GBV100$0.110/hr
wav2vec2-xls-r-300m-ftspeech315MF321.4 GBV100$0.110/hr
romanian-wav2vec2315MF321.4 GBV100$0.110/hr
wav2vec2-large-xlsr-latvian-cv315MF321.4 GBV100$0.110/hr
w2v-xls-r-uk315MF321.4 GBV100$0.110/hr
wav2vec2-xls-r-300m-timit-phoneme315MF321.4 GBV100$0.110/hr
wav2vec2-large-xlsr-kazakh315MF321.4 GBV100$0.110/hr
wav2vec2-xls-r-300m-cs-250315MF321.4 GBV100$0.110/hr
wav2vec2-large-voxrex-swedish315MF321.4 GBV100$0.110/hr
wav2vec2-xls-r-parlaspeech-hr315MF321.4 GBV100$0.110/hr
wav2vec2-large-xls-r-300m-Urdu315MF321.4 GBV100$0.110/hr
wav2vec2-large-xlsr-mvc-swahili315MF321.4 GBV100$0.110/hr
Wav2Vec2-large-xlsr-hindi316MF321.4 GBV100$0.110/hr
wav2vec2-large-xlsr-marathi316MF321.4 GBV100$0.110/hr
wav2vec2-large-xls-r-300m-assamese316MF321.4 GBV100$0.110/hr
wav2vec2-large-xls-r-300m-sinhala-low-LR-part1316MF321.4 GBV100$0.110/hr
wav2vec2-large-xlsr-japanese-hiragana316MF321.4 GBV100$0.110/hr
wav2vec2-large-robust-L2-english-phoneme-recognition316MF321.4 GBV100$0.110/hr
wav2vec2-xls-r-300m-pashto316MF321.4 GBV100$0.110/hr
wav2vec2-large-xlsr-53-amharic316MF321.4 GBV100$0.110/hr
wav2vec2-large-xlsr-korean317MF321.4 GBV100$0.110/hr
my_zh_CN_asr_cv13_model319MF321.4 GBV100$0.110/hr
wav2vec2-conformer-rope-large-960h-ft593MF322.7 GBV100$0.110/hr
Ethio-ASR-multilingual-600M606MF322.7 GBV100$0.110/hr
nb-wav2vec2-1b-bokmaal963MF324.3 GBV100$0.110/hr
nb-wav2vec2-1b-nynorsk963MF324.3 GBV100$0.110/hr
nb-wav2vec2-1b-bokmaal-v2963MF324.3 GBV100$0.110/hr

VRAM is for the precision the model is published in: the weights times a flat 1.2 overhead, with the KV cache not included. See the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.

More from Meta

Submit the job. Everything after that is ours.

Sign up in 60 seconds. Pay for the GPU minutes you actually use.

© 2026 Aquanode. All rights reserved.

All trademarks, logos and brand names are the property of their respective owners.