EuroEval Estonian NLU - Estonian Valence: leaderboard
The "Estonian Valence" subtask of EuroEval Estonian NLU. Estonian-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: Matthews correlation coefficient (%). Source: euroeval.com. Saturation forecast: Estimated already saturated. 251 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Flash Lite (Preview) | 67.41 | #194 |
| 2 | Gemini 3 Flash (Preview) (Thinking) | 66.82 | #82 (Gemini 3 Flash (Preview)) |
| 3 | GPT-5 | 65.74 | #98 |
| 4 | Gemini 2.5 Pro | 64.76 | #152 |
| 5 | Gemma 3 27B (IT) | 64.43 | #523 |
| 6 | Qwen 3 235B A22B 2507 Instruct | 64.41 | #300 |
| 7 | Claude Sonnet 4.6 | 63.92 | #90 |
| 8 | Llama 3.3 70B Instruct | 63.89 | #534 |
| 9 | Qwen 3 235B A22B 2507 FP8 (Thinking) | 63.79 | #246 (Qwen 3 235B A22B 2507 FP8) |
| 10 | Claude Sonnet 4.5 (Thinking) | 63.49 | #142 (Claude Sonnet 4.5) |
| 11 | Gemini 3 Pro (Preview) | 63.19 | #66 |
| 12 | Gemma 3 12B (IT) | 63.17 | #664 |
| 13 | Llama 3.1 70B Instruct | 63.07 | #561 |
| 14 | GPT-5.2 | 63.06 | #110 |
| 15 | Qwen 3 Next 80B A3B Instruct | 63.01 | #368 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-estonian-nlu-estonian-valence · How It Works · Data refreshed daily, snapshot 2026-10-20.