CZ-EVAL - Klokan (Math): leaderboard
The Klokan (Math) subtask of CZ-EVAL (CZ-EVAL) — problems from the Klokan / Math Kangaroo competition posed in Czech; accuracy (%).
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 Preview (1106) | 58.06 | #418 |
| 2 | Claude 3 Opus | 57.81 | #571 |
| 3 | Claude 3 Sonnet | 44.28 | #775 |
| 4 | Mistral Large | 41.08 | #710 |
| 5 | Claude 2.1 | 38.24 | #839 |
| 6 | GPT-3.5 Turbo | 32.47 | #865 |
| 7 | Claude 3 Haiku | 32.23 | #803 |
| 8 | Mistral Medium | 29.4 | #792 |
| 9 | Gemini 1.0 Pro | 28.66 | #933 |
| 10 | Mixtral 8x7B Instruct | 26.69 | #1170 |
| 11 | Command R | 24.85 | #1223 |
| 12 | gemma-7B (IT) | 16.97 | #1496 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=cz-eval-klokan-math · How It Works · Data refreshed daily, snapshot 2026-10-20.