PaliBench - BLEU: leaderboard
PaliBench (2026) asks models to translate 1,700 passages of the Pali Buddhist canon into English, scored against three independent human translations at once. This board reports corpus BLEU on a 0-100 scale; higher is better.
Metric: BLEU (0-100), sacrebleu corpus BLEU, 1,700 Pali canon passages translated into English, each scored against three human reference translations (Bodhi, Sujato, Thanissaro), corpus level, models through OpenRouter; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 63.3 | #119 |
| 2 | Gemini 3 Flash (Preview) | 57.9 | #152 |
| 3 | Claude Opus 4.5 | 56.2 | #160 |
| 4 | DeepSeek V3.2 | 51.3 | #419 |
| 5 | Kimi K2.5 | 47.5 | #266 |
| 6 | GLM-4.7 | 43.7 | #380 |
| 7 | GPT-5.2 | 42.5 | #185 |
| 8 | Qwen 3 235B A22B 2507 Instruct | 40.6 | #616 |
| 9 | Grok 4.1 Fast | 40.1 | #427 |
| 10 | Llama 3.3 70B Instruct | 34.4 | #1385 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=palibench-bleu · How It Works · Data refreshed daily, snapshot 2026-11-21.