Tether AI Research Unveils Compact Translation Models for On-Device Use

Tether AI Research introduced the open TranslatePsy-EuroNano model family for on-device translation between English and nine European languages. Deployment takes 36–89MB versus 633MB for 18 bilingual Firefox Bergamot models, with comparable quality.

• Two multilingual models cover 90 translation directions instead of 18 separate ones
• Deployment size 36–89MB, up to 17.6 times smaller than Firefox counterpart
• BaseQ retains 98.4% of NLLB-200 quality when translating to English
• First CPU translation: 4.2s for TinyQ vs 10.8s for Firefox