SeaEval - Cultural Reasoning - US-Eval (Zero-Shot) — leaderboard

Metric: Accuracy (%). Source: huggingface.co. 25 models tracked.

Top models

#ModelScore
1Qwen 2 72B Instruct87.85
2Llama 3 70B Instruct86.92
3Qwen 2.5 72B Instruct85.98
4Llama 3.1 70B Instruct84.11
5Qwen 2.5 32B Instruct84.11
6Qwen 2.5 14B Instruct82.24
7Gemma 2 9B (IT)81.31
8Qwen 2.5 7B Instruct76.64
9Llama 3.1 8B Instruct72.9
10Qwen 2 7B Instruct72.9
11Llama 3 8B Cpt Sea Lionv2.1 Instruct71.03
12Llama 3 8B Instruct70.09
13Sailor2-8B-Chat70.09
14Gemma 2 2B (IT)69.16
15SeaLLMs-v3-7B-Chat69.16

Interactive version: theaggregate.ai/benchmark?slug=seaeval-cultural-reasoning-us-eval-zero-shot · How the rankings work · Data refreshed daily, snapshot 2026-07-22.