Grip on LLMs - Dutch Factuality Average: leaderboard

Metric: Accuracy (%; mean of three GP-IRT estimates, Dutch machine translation). Source: arxiv.org. 31 models tracked.

Top models

#ModelScore
1GPT-576
2Mistral Medium 375
3GPT-4o73
4Qwen 3 32B72
5Mistral Large 371
6Mistral Small 370
7GPT-5 Mini69
8Gemma 3 27B (IT)68
9DeepSeek R1 Distill Qwen 32B67
10GPT-4o Mini65
11aya-expanse-32B62
12Gemma 3 12B (IT)61
13GPT-5 Nano61
14Qwen 3 8B59
15GPT-OSS-20B57

Interactive version: theaggregate.ai/benchmark?slug=grip-on-llms-dutch-factuality-average · How It Works · Data refreshed daily, snapshot 2026-09-19.