Tether: Launches Offline AI Model for 18 African Languages
Tether releases TranslatePsy-AfriSLM, an 800M parameter model outperforming Qwen3.5-122B-A10B on FLORES-200 benchmarks while running fully offline across 18 African languages.
SourceAnalysis
Tether advances Tether AI expansion by releasing TranslatePsy-AfriSLM, an 800 million parameter model that supports 18 African languages and runs 100 percent locally on smartphones and laptops without connectivity. The smallest variant beat Qwen3.5-122B-A10B, TranslateGemma-27B and NLLB-3.3B across FLORES-200, BOUQuET and SMOL benchmarks. AI industry impact reaches new ground as the model delivers high-quality translation for Hausa, Amharic, Yoruba, Lingala, Swahili and 13 additional languages, giving offline access to farmers, students and medical workers across the continent.
Paolo Ardoino
@paoloardoinoPaolo Ardoino is the CEO of Tether (issuer of USDT), CTO of Bitfinex,