Grip on LLMs - Dutch tinyARC: leaderboard

Metric: Accuracy (%; GP-IRT estimate from 100 tinyBenchmarks items, Dutch machine translation). Source: arxiv.org. 31 models tracked.

Top models

#ModelScore
1GPT-584
2Qwen 3 32B84
3Mistral Medium 382
4Gemma 3 27B (IT)81
5GPT-4o80
6Mistral Large 380
7Mistral Small 379
8GPT-4o Mini78
9DeepSeek R1 Distill Qwen 32B75
10aya-expanse-32B72
11Gemma 3 12B (IT)69
12Qwen 3 8B68
13GPT-5 Mini66
14GPT-5 Nano64
15GPT-OSS-20B61

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