Translation en to Set1 COMET22: leaderboard
COMET-22 is an ensemble machine translation evaluation metric combining a COMET estimator model trained with Direct Assessments and a multitask model that predicts sentence-level scores and word-level OK/BAD tags. It demonstrates improved correlations compared to state-of-the-art metrics and increased robustness to critical errors. (Source: benchmarklist.com, self-reported.)
Metric: en→Set1 COMET22 (imported). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.5 Sonnet | 89.4 | #350 |
| 2 | GPT-4o | 89.2 | #343 |
| 3 | Gemini 1.5 Pro (002) | 89.1 | #349 |
| 4 | Nova Pro | 89.1 | #569 |
| 5 | Nova Lite | 88.8 | #754 |
| 6 | GPT-4o Mini | 88.7 | #599 |
| 7 | Gemini 1.5 Flash (002) | 88.5 | #583 |
| 8 | Claude 3.5 Haiku | 88.5 | #568 |
| 9 | Nova Micro | 88.5 | #898 |
| 10 | Llama 3.2 90B | 88.2 | #596 |
| 11 | Gemini 1.5 Flash-8B (001) | 88 | #791 |
| 12 | Llama 3.2 11B | 85.7 | #1199 |
| 13 | Llama 3.1 8B | 85.5 | #1151 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=translation-en-to-set1-comet22 · How It Works · Data refreshed daily, snapshot 2026-10-20.