EuroEval Bulgarian NLU - ScaLA BG: leaderboard
Metric: Linguistic acceptability Score (%). Source: euroeval.com. 212 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 83.73 |
| 2 | Gemini 3.1 Flash Lite (Preview) | 83.09 |
| 3 | GPT-5.4 Mini (High) | 77.4 |
| 4 | GPT-5.2 | 77.04 |
| 5 | Gemini 2.5 Flash (Thinking) | 76.7 |
| 6 | Claude Sonnet 4.6 | 75.49 |
| 7 | GPT-4.1 | 73.45 |
| 8 | Gemini 2.5 Pro | 72.75 |
| 9 | GPT-5.4 Mini (Medium) | 71.87 |
| 10 | Grok 4.20 Beta (0309) (Reasoning) | 70.6 |
| 11 | Gemma 4 31B (IT) | 68.85 |
| 12 | GPT-5 Mini | 68.28 |
| 13 | GPT-5.4 Nano (Medium) | 68.2 |
| 14 | Claude Sonnet 4.5 (Thinking) | 67.95 |
| 15 | GPT-5 | 67.3 |
Interactive version: theaggregate.ai/benchmark?slug=euroeval-bulgarian-nlu-scala-bg · How It Works · Data refreshed daily, snapshot 2026-09-05.