Tether addresses AI underinvestment in Africa with open-source machine translation models
800M-parameter model beats Qwen's 122B on African languages. Tether's TranslatePsy-AfriSLM just changed what's possible in low-resource translation.
Why it matters
A major open-weight model release addressing a real capability gap: small, efficient translation models that outperform much larger systems on 19 African languages, with no cloud dependence. This is frontier capability work solving a neglected problem.
The key facts
8 to knowTranslatePsy-AfriSLM: 800M parameters outperforms Qwen3.5-122B-A10B, TranslateGemma-27B, NLLB-3.3B on benchmarks
Covers 19 Sub-Saharan African languages: Hausa, Amharic, Yoruba, Lingala, Swahili, Igbo, Zulu, Somali, Oromo, Malagasy, Kinyarwanda, Xhosa, Afrikaans, Wolof, Luganda, Nyanja, Shona, Tswana, Southern Sotho
TranslatePsy-EuroNano: 17.6x smaller than comparable European baseline while maintaining translation quality
Runs on-device, no cloud API required; fully open source on Hugging Face
UNESCO: AI could generate $1.2 trillion for Africa's economy by 2030 (6% of GDP)
Only one African country (South Africa) scores 50+ out of 100 on 2025 Government AI Readiness Index
Available through QVAC SDK for Android, iOS, Linux, macOS, Windows
Use cases: healthcare translation, agricultural information, humanitarian response, offline capability
Go to the source
CIOcio.com
Publisher excerpt: Most translation models are primarily trained for high-resource Asian and European languages. Most African languages, spoken by hundreds of millions of people, are relatively neglected, compared to their high-resource counterparts. Although AI underinvestment across the African continent has…