WMT25 English-Maasai ESA: leaderboard
WMT25 General Translation Task English-to-Maasai ESA human-evaluation score for selected submitted systems.
Metric: ESA Human Score. Source: github.com. Saturation forecast: Around 2031. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 9.8 | #152 |
| 2 | aya-expanse-8B | 6 | #931 |
| 3 | Llama 4 Maverick | 5.2 | #464 |
| 4 | aya-expanse-32B | 3.1 | #692 |
| 5 | DeepSeek V3 | 3 | #321 |
| 6 | Llama 3.1 8B | 3 | #1151 |
| 7 | Gemma 3 12B | 3 | #671 |
| 8 | Qwen 2.5 7B | 2.8 | #922 |
| 9 | Qwen 3 235B A22B | 2.7 | #313 |
| 10 | Command A | 1.3 | #508 |
| 11 | EuroLLM-9B | 0.7 | #1254 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=wmt25-english-maasai-esa · How It Works · Data refreshed daily, snapshot 2026-10-20.