OpenHuEval: leaderboard
OpenHuEval evaluates large language models on Hungarian-specific tasks, including real user queries, self-awareness, proverb reasoning, generative evaluation, and fill-in-the-blank tasks.
Metric: Macro Average (computed) (imported). Source: github.com. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 63.77 | #343 |
| 2 | DeepSeek R1 | 62.31 | #251 |
| 3 | DeepSeek V3 | 57.1 | #321 |
| 4 | Llama 3.1 70B Instruct | 50.41 | #561 |
| 5 | O1 Mini | 49.38 | #323 |
| 6 | GPT-4o Mini | 49.33 | #599 |
| 7 | Llama 3.1 8B Instruct | 28.73 | #1035 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openhueval · How It Works · Data refreshed daily, snapshot 2026-10-20.