Polish Knowledge Benchmark - History: leaderboard

Metric: Accuracy (%). Source: huggingface.co. 68 models tracked.

Top models

#ModelScore
1Gemini 3 Flash (Preview)94.41
2Gemini 3.6 Flash93.74
3DeepSeek V4 Flash93.06
4Gemini 3.1 Flash Lite (Preview)92.47
5Gemini 3.1 Flash Lite92.39
6Mistral Large 391.99
7Qwen 3.5 397B A17B91.48
8Grok 4.1 Fast91.12
9DeepSeek V3.290.92
10Gemini 2.5 Flash90.84
11Gemini 3.5 Flash Lite90.64
12Gemini 2.0 Flash (001)90.33
13Mistral Medium 3.190.21
14DeepSeek V3 (0324)90.05
15Gemma 4 31B (IT)89.89

Interactive version: theaggregate.ai/benchmark?slug=polish-knowledge-benchmark-history · How It Works · Data refreshed daily, snapshot 2026-09-19.