ResearchBenchmarksMachine Translation on WMT document-level En→Ja 24 (test)Follow4.373M-Prometheus Scorepfnet/plamo-2-translate1.369482.149242.9293.70876Jun 19, 2026Evaluation ResultsMethodMethodLinksM-Prometheus ScoreAverage Scorepfnet/plamo-2-translate2026.064.3734.393tokyotech-llm/Llama-3.1-Swallow-8B-Instruct-v0.52026.064.2493.049shisa-ai/shisa-v2.1-llama3.2-3b2026.064.1543.486tencent/HY-MT1.5-7B2026.063.9533.673shisa-ai/shisa-v2.1-lfm2-1.2b2026.063.843.103google/translategemma-12b-it2026.063.7283.285CAT-Translate-1.4B2026.063.3963.191Qwen/Qwen2.5-14B-Instruct2026.063.3433.106google/translategemma-4b-it2026.063.1072.872nvidia/NVIDIA-Nemotron-Nano-9B-v2-Japanese2026.063.0893.248Unbabel/Tower-Plus-9B2026.063.0533.311CAT-Translate-7B2026.062.9293.143tencent/HY-MT1.5-1.8B2026.062.9112.847microsoft/phi-42026.062.8583.187mistralai/Ministral-8B-Instruct-24102026.062.8522.676CAT-Translate-3.3B2026.062.8283.057CAT-Translate-0.8B2026.062.8172.745sbintuitions/sarashina2.2-3b-instruct-v0.12026.062.7342.715SakanaAI/TinySwallow-1.5B-Instruct2026.062.5562.49LiquidAI/LFM2.5-1.2B-JP2026.062.5382.399google/gemma-2-2b-jpn-it2026.062.3432.37LiquidAI/LFM2-350M-ENJP-MT2026.062.0362.362llm-jp/llm-jp-3.1-1.8b-instruct42026.061.9822.291Rakuten/RakutenAI-2.0-mini-instruct2026.061.8221.804meta-llama/Llama-3.2-3B-Instruct2026.061.4851.667