JP-TL-Bench - Japanese to English - Hard: leaderboard
Hard Japanese-to-English items of JP-TL-Bench (Lin and Lensenmayer, Shisa.AI, 2026): 21 Japanese passages including classic literary prose, light-novel fiction, essays, news and recipes, translated into English and compared pairwise and reference-free against the frozen Base Set v1.0 anchor translations by gemini-2.5-flash. Bradley-Terry strength fitted on this slice alone, mapped to a 0-10 LT score; higher is better.
Metric: LT Score (0-10; Hard items only; anchored pairwise Bradley-Terry, gemini-2.5-flash judge). Source: arxiv.org. Saturation forecast: Estimated already saturated. 64 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 9.99 | #262 |
| 2 | Gemini 3 Flash (Preview) | 9.99 | #141 |
| 3 | Gemini 3 Pro (Preview) | 9.98 | #111 |
| 4 | Qwen 3 235B A22B 2507 Instruct | 9.97 | #619 |
| 5 | GPT-OSS-120B | 9.95 | #699 |
| 6 | DeepSeek V3.1 Terminus | 9.93 | #442 |
| 7 | Kimi K2 0905 | 9.92 | #555 |
| 8 | Ministral-3-14B-Instruct-2512 | 9.89 | #1595 |
| 9 | Gemini 2.5 Flash | 9.88 | #452 |
| 10 | Qwen 3 30B A3B 2507 Instruct | 9.84 | #1078 |
| 11 | Gemma 3 27B (IT) | 9.77 | #1306 |
| 12 | GPT-4o (2024-08-06) | 9.74 | #731 |
| 13 | Ling-flash-2.0 | 9.19 | #1352 |
| 14 | ABEJA-Qwen2.5-32B-Japanese-v1.0 | 9.08 | #1304 |
| 15 | ELYZA-Shortcut-1.0-Qwen-32B | 8.94 | #1420 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=jp-tl-bench-japanese-to-english-hard · How It Works · Data refreshed daily, snapshot 2026-11-12.