HardMTBench - English to Chinese: leaderboard

HardMTBench (2026) is a difficulty-aware Chinese-English translation test of 10,000 pairs from 12 knowledge-intensive domains, selected for domain-knowledge density and translation difficulty. This board scores the English-to-Chinese direction with GEMBA-DA, an LLM direct assessment from 0 to 100; higher is better.

Metric: GEMBA-DA direct assessment score (0-100) of the English-to-Chinese translations, judged by gpt-oss-120b, on HardMTBench, 10,000 domain-balanced, difficulty-selected Chinese-English pairs from 12 knowledge-intensive domains, each translated in both directions under one unified protocol; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3.1 Pro (Preview)92.78#99
2GPT-5.592.66#54
3Qwen 3.5 397B A17B92.13#264
4DeepSeek V4 ProThinking91.79#196
5Qwen 3.6 35B A3B91.05#481
6Qwen-MT-Plus90.21
7Hy-MT2-30B-A3B89.71
8Gemma 4 31B (IT)Thinking89.03#350
9gemma-4-26B-A4B-itThinking88.53
10Gemma 4 26B A4B (IT)87.57#556
11Hy-MT2-7B86.91
12Gemma 4 E4B (IT)Thinking83.24
13gemma-4-E4B-it82.82#1539
14Hy-MT2-1.8B82.28
15Gemma 4 E2B (IT)Thinking80.87#2451

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

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