EuroEval Hungarian Summarization - Hunsum: leaderboard
Metric: Score (%). Source: euroeval.com. 149 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 | 34.8 |
| 2 | Gemini 3 Pro (Preview) | 34.53 |
| 3 | Apertus-70B-Instruct-2509 | 33.96 |
| 4 | Mistral Small 3.1 | 33.66 |
| 5 | Gemma 3 27B (IT) | 33.49 |
| 6 | Mistral Small 3.2 | 33.46 |
| 7 | Gemma 4 31B (IT) | 33.42 |
| 8 | GPT-5.4 Mini (High) | 33.39 |
| 9 | Claude Sonnet 4.5 (Non-reasoning) | 33.37 |
| 10 | GPT-5.4 Mini (Medium) | 33.3 |
| 11 | Gemma 3 12B (IT) | 33.26 |
| 12 | Gemma 3n E4B (IT) | 33.2 |
| 13 | Gemini 3.1 Flash Lite (Preview) | 33.19 |
| 14 | Qwen 3 30B A3B 2507 Instruct | 33.15 |
| 15 | Claude Sonnet 4.5 (Thinking) | 32.91 |
Interactive version: theaggregate.ai/benchmark?slug=euroeval-hungarian-summarization-hunsum · How It Works · Data refreshed daily, snapshot 2026-09-05.