Kaggle MultiLoKo - Russian: leaderboard

Metric: Exact Match (%). Source: www.kaggle.com. 48 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)78.31
2Gemini 3.5 Flash74
3Gemini 3 Flash (Preview)69.2
4Gemini 2.5 Pro (Preview 06-05)64.4
5Gemini 3.1 Flash Lite (Preview)48.8
6O148
7GPT-546.4
8Gemini 2.5 Flash (Preview 05-20)45.2
9O343.6
10GPT-5.442.8
11Claude 3.5 Sonnet (20241022)41.6
12Claude 3.7 Sonnet41.6
13Gemini 1.5 Pro (002)38.4
14Claude Opus 4.138.4
15Claude Opus 438

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