EuroEval Estonian NLU: leaderboard
Estonian-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: NLU Average Score (%). Source: euroeval.com. Saturation forecast: Around January 2027. 258 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) (Thinking) | 62.58 | #82 (Gemini 3 Flash (Preview)) |
| 2 | Gemini 3 Pro (Preview) | 61.56 | #66 |
| 3 | Gemini 3.1 Flash Lite (Preview) | 60.4 | #194 |
| 4 | Gemini 3 Flash (Preview) (Non-reasoning) | 60.32 | #82 (Gemini 3 Flash (Preview)) |
| 5 | Gemini 2.5 Pro | 60.31 | #152 |
| 6 | GPT-5 | 59.94 | #98 |
| 7 | Claude Sonnet 4.6 | 58.66 | #90 |
| 8 | GPT-5.4 Mini (2026-03-17) (High) | 58.38 | #169 (GPT-5.4 Mini (2026-03-17)) |
| 9 | Gemma 4 31B (IT) | 57.54 | #176 |
| 10 | GPT-5.2 | 57.12 | #110 |
| 11 | Gemini 2.5 Flash (Thinking) | 57.05 | #245 (Gemini 2.5 Flash) |
| 12 | Claude Sonnet 4.5 (Thinking) | 56.54 | #142 (Claude Sonnet 4.5) |
| 13 | Qwen 3 235B A22B 2507 Instruct | 56.44 | #300 |
| 14 | Gemma 3 27B (IT) | 56.01 | #523 |
| 15 | gemma-3-27B-pt | 55.45 | #502 |
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 · How It Works · Data refreshed daily, snapshot 2026-10-20.