CanMT - English to Japanese: leaderboard

CanMT (2026) is a culture-aware machine translation benchmark of novel sentences containing culture-specific items, scored by an LLM judge with references on five translation-quality dimensions. This board is the translation score on a 1 to 7 scale for English to Japanese; higher is better.

Metric: Translation score (1-7) on the 103 English-to-Japanese CanMT sentences containing culture-specific items, GPT-5-nano judge with the reference translation scoring contextual accuracy, cultural adaptation, functional equivalence, fidelity and naturalness on 7-point Likert scales, averaged per sentence; default translation prompt; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 22 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Grok 4.15.34#472
2DeepSeek V3.25.21#425
3GPT-45.16#988
4DeepSeek R15.08#484
5Gemini 2.5 Flash Lite5.08#972
6GPT-4o4.99#754
7Qwen 2.5 72B Instruct4.84#1124
8Llama 3.3 70B Instruct4.83#1394
9Seed-X-PPO-7B4.69
10Qwen 3 14BNon-reasoning4.65#1284
11Google Translate4.64
12Qwen 3 32BNon-reasoning4.47#983
13Qwen 2.5 32B Instruct4.4#1253
14Qwen 3 8BNon-reasoning4.25#1890
15Qwen 2.5 14B Instruct4.08#1788

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=canmt-english-to-japanese · How It Works · Data refreshed daily, snapshot 2026-11-22.