CanMT - Japanese to English: 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 Japanese to English; higher is better.

Metric: Translation score (1-7) on the 116 Japanese-to-English 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
1GPT-45.18#970
2Grok 4.15.04#462
3GPT-4o5.03#751
4DeepSeek R14.97#482
5Gemini 2.5 Flash Lite4.96#964
6DeepSeek V3.24.88#419
7Qwen 2.5 72B Instruct4.67#1117
8Qwen 2.5 32B Instruct4.6#1247
9Llama 3.3 70B Instruct4.44#1385
10Qwen 3 32BNon-reasoning4.39#972
11Google Translate4.35
12Qwen 2.5 14B Instruct4.32#1775
13Qwen 3 14BNon-reasoning4.24#1269
14Seed-X-PPO-7B4.21
15Qwen 3 8BNon-reasoning4.02#1873

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=canmt-japanese-to-english · How It Works · Data refreshed daily, snapshot 2026-11-21.