EuroEval Czech NLU - CS GEC: leaderboard
The "CS GEC" subtask of EuroEval Czech NLU. Czech-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: Around January 2027. 229 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) (Thinking) | 64.64 | #82 (Gemini 3 Flash (Preview)) |
| 2 | Gemini 2.5 Pro | 64.37 | #152 |
| 3 | Gemini 2.5 Flash (Thinking) | 63.31 | #245 (Gemini 2.5 Flash) |
| 4 | Claude Sonnet 4.6 | 60.22 | #90 |
| 5 | Gemini 3.1 Flash Lite (Preview) | 59.64 | #194 |
| 6 | GPT-4.1 | 59.12 | #248 |
| 7 | Gemini 3 Flash (Preview) (Non-reasoning) | 59.08 | #82 (Gemini 3 Flash (Preview)) |
| 8 | Qwen 3.5 35B A3B Base | 58.42 | #310 |
| 9 | GPT-5.4 Mini (2026-03-17) (High) | 58.1 | #169 (GPT-5.4 Mini (2026-03-17)) |
| 10 | Gemini 2.5 Flash Lite (Thinking) | 57.87 | #422 (Gemini 2.5 Flash Lite) |
| 11 | Qwen 3.5 27B FP8 | 56.09 | #164 |
| 12 | Gemma 4 26B A4B (IT) | 55.73 | #297 |
| 13 | Gemini 3 Pro (Preview) | 55.12 | #66 |
| 14 | GPT-4.1 Mini | 53.79 | #355 |
| 15 | Qwen 3.5 35B A3B FP8 | 52.44 | #362 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-czech-nlu-cs-gec · How It Works · Data refreshed daily, snapshot 2026-10-20.