Skip to content

Benchmark

SeaLLMs SeaExam public (Thai)

Multilingual benchmarks compare model performance across languages. Scores are percentages. Higher is better.

Models measured
57
Scores recorded
57
Sources
1
IQ weight
1.42

Score distribution

59.6%
Mean
63.2%
Median
15.1%
Std dev
39.7%
10th percentile
77.8%
90th percentile
26.2%
Min
26.2%92.9%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.42 for chat models, measured over 57 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 2.5 Pro Preview 03-25· SeaLLMs Leaderboard for SEA Languages data docs92.9%
2Claude 3.5 Sonnet· SeaLLMs Leaderboard for SEA Languages data docs80.9%
3Claude Sonnet 3.7· SeaLLMs Leaderboard for SEA Languages data docs80.9%
4Gemini 2.0 Flash· SeaLLMs Leaderboard for SEA Languages data docs79%
5gemini-pro-1.5· SeaLLMs Leaderboard for SEA Languages data docs78%
6GPT 4.1· SeaLLMs Leaderboard for SEA Languages data docs78%
7DeepSeek V3· SeaLLMs Leaderboard for SEA Languages data docs77.8%
8GPT 4o· SeaLLMs Leaderboard for SEA Languages data docs77%
9Qwen2.5 72B· SeaLLMs Leaderboard for SEA Languages data docs76.3%
10Llama 4 Maverick· SeaLLMs Leaderboard for SEA Languages data docs75.8%
11Qwen2.5 72B Instruct· SeaLLMs Leaderboard for SEA Languages data docs75.2%
12Qwen 2 (72B)· SeaLLMs Leaderboard for SEA Languages data docs74.4%
13Gemini 2.0 Flash Lite· SeaLLMs Leaderboard for SEA Languages data docs74.2%
14GPT-4.1 mini· SeaLLMs Leaderboard for SEA Languages data docs72.2%
15Qwen2 72B Instruct· SeaLLMs Leaderboard for SEA Languages data docs71.7%
16Llama 3.1 70B Instruct· SeaLLMs Leaderboard for SEA Languages data docs70.6%
17Qwen 2.5 14B Instruct· SeaLLMs Leaderboard for SEA Languages data docs68.8%
18Qwen2.5 14B· SeaLLMs Leaderboard for SEA Languages data docs68.7%
19Gemini 1.5 Flash· SeaLLMs Leaderboard for SEA Languages data docs68.3%
20Gemma 3 27B Pt· SeaLLMs Leaderboard for SEA Languages data docs68%
21Gemma 3 27B IT· SeaLLMs Leaderboard for SEA Languages data docs67.5%
22Meta Llama 3 70b· SeaLLMs Leaderboard for SEA Languages data docs65.9%
23Llama 3.1 70B· SeaLLMs Leaderboard for SEA Languages data docs65.8%
24GPT-4o mini· SeaLLMs Leaderboard for SEA Languages data docs65.7%
25Llama 3 70B Instruct· SeaLLMs Leaderboard for SEA Languages data docs65.3%
26Gemma 2 27B IT· SeaLLMs Leaderboard for SEA Languages data docs64.6%
27Claude Haiku 3.5· SeaLLMs Leaderboard for SEA Languages data docs64.4%
28Llama 4 Scout· SeaLLMs Leaderboard for SEA Languages data docs64.4%
29Gemma 3 12B IT· SeaLLMs Leaderboard for SEA Languages data docs63.2%
30gemma2-27b· SeaLLMs Leaderboard for SEA Languages data docs61.9%
31Qwen2.5 7B Instruct· SeaLLMs Leaderboard for SEA Languages data docs61.6%
32Qwen2.5 7B· SeaLLMs Leaderboard for SEA Languages data docs60.9%
33Gemma 2 9B IT· SeaLLMs Leaderboard for SEA Languages data docs60.1%
34GPT-4.1 nano· SeaLLMs Leaderboard for SEA Languages data docs57.1%
35Claude 3 Haiku· SeaLLMs Leaderboard for SEA Languages data docs56.4%
36Qwen2 7B Instruct· SeaLLMs Leaderboard for SEA Languages data docs56%
37gemma-2-9b· SeaLLMs Leaderboard for SEA Languages data docs55%
38Qwen2.5 3B Instruct· SeaLLMs Leaderboard for SEA Languages data docs52.7%
39Llama 3.1 8B Instruct· SeaLLMs Leaderboard for SEA Languages data docs51%
40GLM-4-9b-chat· SeaLLMs Leaderboard for SEA Languages data docs50.7%
41Llama 3.1 8B· SeaLLMs Leaderboard for SEA Languages data docs49.8%
42GLM 4 9B 0414· SeaLLMs Leaderboard for SEA Languages data docs49.7%
43Gemma 3 4B IT· SeaLLMs Leaderboard for SEA Languages data docs49.3%
44Llama 3 8B Instruct· SeaLLMs Leaderboard for SEA Languages data docs47.8%
45Meta Llama 3 8b· SeaLLMs Leaderboard for SEA Languages data docs47.7%
46Qwen2.5 1.5B Instruct· SeaLLMs Leaderboard for SEA Languages data docs43.6%
47Llama 3.2 3B· SeaLLMs Leaderboard for SEA Languages data docs42.9%
48Qwen2.5 1.5B· SeaLLMs Leaderboard for SEA Languages data docs42.8%
49Gemma 2 2B IT· SeaLLMs Leaderboard for SEA Languages data docs42.4%
50Mixtral 8x7B Instruct v0.1· SeaLLMs Leaderboard for SEA Languages data docs42.2%

Showing 1–50 of 57 results.

Data credits: SeaExam and SeaBench leaderboard, SeaLLMs (DAMO-NLP-SG). Published leaderboard scores; results licence not stated.