Kaggle MultiLoKo - Simplified Mandarin: leaderboard

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

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)70.04
2Gemini 3.5 Flash62.8
3Gemini 3 Flash (Preview)59.6
4Gemini 2.5 Pro (Preview 06-05)53.6
5Gemini 3.1 Flash Lite (Preview)49.2
6DeepSeek R1 052847.6
7GPT-5.443.6
8Gemini 2.5 Flash (Preview 05-20)43.2
9Gemini 2.0 Flash42.4
10DeepSeek V341.6
11Claude Opus 441.2
12O140
13Claude Opus 4.140
14GPT-537.2
15O337.2

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