EuroEval Faroese NLU - FoSent: leaderboard
Metric: Sentiment classification Score (%). Source: euroeval.com. 400 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 91.61 |
| 2 | Gemini 2.5 Flash (Preview 04-17) | 84.89 |
| 3 | Gemini 3 Pro (Preview) | 83.1 |
| 4 | O3 (2025-04-16) | 81 |
| 5 | Claude Sonnet 4.6 | 77.8 |
| 6 | GPT-4.1 | 77.78 |
| 7 | GPT-5 Mini | 77.66 |
| 8 | Gemma 4 31B (IT) | 76.96 |
| 9 | Gemini 3.1 Flash Lite (Preview) | 75.3 |
| 10 | Claude Sonnet 4.5 (Non-reasoning) | 74.64 |
| 11 | Claude Sonnet 4.5 (Thinking) | 74.33 |
| 12 | GPT-5.4 Mini (High) | 73.95 |
| 13 | Grok 2 (1212) | 73.84 |
| 14 | GPT-5.4 Mini (Medium) | 73.46 |
| 15 | GPT-5 Mini (High) | 72.91 |
Interactive version: theaggregate.ai/benchmark?slug=euroeval-faroese-nlu-fosent · How It Works · Data refreshed daily, snapshot 2026-09-05.