Tether AI Research has launched open-source translation models designed to run locally on smartphones, laptops and other edge devices, with a focus on improving access to digital information across Sub-Saharan Africa.
The release includes QVAC TranslatePsy-AfriSLM, which supports 19 African languages, and QVAC TranslatePsy-AfriNano, which supports eight. A parallel model, QVAC TranslatePsy-EuroNano, covers nine European languages.
Because the models process translations on the device, they can work without an internet connection and keep users’ data away from third-party cloud servers.
Offline AI translation for underserved communities
Language remains a significant barrier to modern AI access for hundreds of millions of people across Africa. Many of the most powerful tools support only a small number of major languages and rely on cloud connectivity, limiting their usefulness in communities with unreliable or expensive internet access.
TranslatePsy-AfriSLM supports Hausa, Amharic, Yoruba, Lingala, Swahili, Igbo, Zulu, Somali, Oromo, Malagasy, Kinyarwanda, Xhosa, Afrikaans, Wolof, Luganda, Nyanja, Shona, Tswana and Southern Sotho. Together, these languages span West, East, Central and Southern Africa and represent roughly half of the continent’s population.
Translation across 19 African languages could help deliver courses, educational resources, scientific material and AI-powered learning tools directly to children and adults in their own languages.
Despite containing just 800 million parameters, the smallest TranslatePsy-AfriSLM model outperformed Qwen3.5-122B-A10B, TranslateGemma-27B and NLLB-3.3B across the FLORES-200, BOUQuET and SMOL translation benchmarks.
Tether AI Research said a new quality-estimation filtering method removes up to 96% of low-quality open-source training data. Improving the quality of the underlying data enabled the company to achieve stronger translation performance with significantly smaller models.
Potential applications in healthcare and education
Healthcare is one of the potential high-impact applications. Patients may speak different local languages, while connectivity can be unreliable in the communities that most need access to trustworthy information.
Combined with Tether QVAC MedPsy, a small foundation model for medical and healthcare applications, TranslatePsy-AfriSLM could help deliver medical knowledge and health education in local languages to hundreds of millions of people.
Such systems would require appropriate safeguards and clear boundaries between health education and clinical care, but their potential impact is substantial.
Agriculture, humanitarian response and cross-border communication
The potential uses extend beyond education and healthcare. Farmers could receive agricultural information in local languages, helping convert technical knowledge into practical guidance.
In humanitarian and disaster-response settings, local-language translation could support communication in camps and affected areas where connectivity is limited. For NGOs and field organizations, the technology could help workers communicate across multiple communities without carrying separate translation systems for every language.
Tether said it has spent years building physical touchpoints in these communities. Across Sub-Saharan Africa, its solar-powered kiosks allow residents to charge a phone, swap a battery and access digital financial services in areas beyond the reach of the electricity grid and banking system.
The company said those same hubs could also become access points for local-language educational content, allowing families to charge a phone while children watch a science documentary or parents learn new farming techniques.
TranslatePsy-EuroNano supports nine European languages
The same design approach underpins Tether AI Research’s parallel release for European languages. TranslatePsy-EuroNano replaces dozens of separate bilingual models with two compact multilingual models for each performance tier.
Using English as a pivot language, the models support 90 translation directions across nine European languages. The smallest deployment requires 36MB of storage, compared with 633MB for an equivalent Firefox offline translation configuration, reducing storage requirements by approximately 94%.
TranslatePsy-EuroNano, the highest-quality model in the European family, retained 98.4% of Meta’s NLLB-200 translation quality when translating into English while using a fraction of the storage required by larger systems.
“Four billion people were left behind by the traditional financial system, and the most powerful technology of our age has repeated that failure,” said Paolo Ardoino, CEO of Tether. “Language should not determine who can benefit from artificial intelligence. Open translation models like these are a step toward a future where education and AI tools reach hundreds of millions of people who have neither reliable connectivity nor access to expensive systems. A mother could get real medical information she understands, instead of guessing. A child could learn in their own language. That is the future we are building through QVAC.”
Model availability and research
TranslatePsy-AfriSLM is available for download on Hugging Face in three sizes, including full-precision and smaller quantized versions:
qvac/TranslatePsy-AfriSLM-0.8Bqvac/TranslatePsy-AfriSLM-2Bqvac/TranslatePsy-AfriSLM-4B
TranslatePsy-Nano models are available through the Hugging Face collection. The collection includes full-precision and quantized versions of models supporting European and African language translation:
qvac/TranslatePsy-EuroNanoqvac/TranslatePsy-AfriNano
The research behind TranslatePsy-AfriSLM has also been accepted for presentation at the Empirical Methods in Natural Language Processing (EMNLP) 2026 conference.
About Tether AI Research and QVAC
Tether AI Research is part of Tether’s broader effort to advance freedom, transparency and innovation through technology. Its mission is to help people and organizations connect and share information directly without unnecessary intermediaries.
By creating secure, peer-to-peer systems, Tether AI Research aims to give users greater control over their data, communications and digital interactions. The initiative seeks to replace centralized models with decentralized infrastructure designed for privacy, efficiency and resilience.
References to Tether AI Research mean Tether Data, S.A. de C.V.
QVAC is Tether’s AI research initiative focused on developing open, decentralized and adaptive intelligence systems. Its mission is “Local AI and Infinite Intelligence,” with the goal of enabling AI to live and learn on any device while empowering individuals and communities rather than concentrating power in corporate data centers.
