iPhone

January 16, 2026, 11:06 PM EST. TranslateGemma unveils a suite of open translation models built on Gemma 3, offered in 4B, 12B, and 27B sizes to cover 55 languages. The project distills knowledge from larger models into compact, high-performance open models, delivering efficiency without quality compromise. In tests, the 12B TranslateGemma model outperforms the Gemma 3 27B baseline on WMT24++ benchmark using MetricX. For developers, this means high-fidelity translation with less than half the parameters of the baseline, enabling higher throughput and lower latency while preserving accuracy. The 4B model rivals the larger 12B baseline, making it suitable for mobile inference. Evaluations on WMT24++ across 55 languages-spanning high, mid, and low-resource languages-show reduced error rates relative to the Gemma baseline.