Machine Translation on WMT Ja→En 21 (test) benchmark leaderboard · SOTA2 Research

Machine Translation on WMT Ja→En 21 (test)

4.413M-Prometheus Scorepfnet/plamo-2-translate

Evaluation Results

Method
pfnet/plamo-2-translate2026.06	4.413	4.393
Unbabel/Tower-Plus-9B2026.06	3.568	3.311
microsoft/phi-42026.06	3.515	3.187
nvidia/NVIDIA-Nemotron-Nano-9B-v2-Japanese2026.06	3.407	3.248
tencent/HY-MT1.5-7B2026.06	3.394	3.673
CAT-Translate-7B2026.06	3.356	3.143
CAT-Translate-3.3B2026.06	3.286	3.057
CAT-Translate-1.4B2026.06	2.985	3.191
Qwen/Qwen2.5-14B-Instruct2026.06	2.868	3.106
google/translategemma-12b-it2026.06	2.842	3.285
shisa-ai/shisa-v2.1-llama3.2-3b2026.06	2.818	3.486
tencent/HY-MT1.5-1.8B2026.06	2.783	2.847
sbintuitions/sarashina2.2-3b-instruct-v0.12026.06	2.696	2.715
LiquidAI/LFM2-350M-ENJP-MT2026.06	2.689	2.362
CAT-Translate-0.8B2026.06	2.672	2.745
google/translategemma-4b-it2026.06	2.636	2.872
llm-jp/llm-jp-3.1-1.8b-instruct42026.06	2.6	2.291
mistralai/Ministral-8B-Instruct-24102026.06	2.501	2.676
SakanaAI/TinySwallow-1.5B-Instruct2026.06	2.424	2.49
google/gemma-2-2b-jpn-it2026.06	2.398	2.37
shisa-ai/shisa-v2.1-lfm2-1.2b2026.06	2.366	3.103
LiquidAI/LFM2.5-1.2B-JP2026.06	2.26	2.399
meta-llama/Llama-3.2-3B-Instruct2026.06	1.85	1.667
tokyotech-llm/Llama-3.1-Swallow-8B-Instruct-v0.52026.06	1.849	3.049
Rakuten/RakutenAI-2.0-mini-instruct2026.06	1.785	1.804