Skip to content

Benchmark

SeaLLMs SeaBench private (Thai)

Multilingual benchmarks compare model performance across languages. Higher scores are better.

Models measured
32
Scores recorded
32
Sources
1
IQ weight
1.84

Score distribution

7.2
Mean
7.5
Median
1.4
Std dev
5.4
10th percentile
8.6
90th percentile
2.7
Min
2.78.9
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.84 for chat models, measured over 32 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT 4.1judge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.9
2DeepSeek V3judge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.7
3Claude Sonnet 3.7judge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.7
4Gemini 2.0 Flashjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.6
5GPT 4ojudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.5
6Claude 3.5 Sonnetjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.5
7gemini-pro-1.5judge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.5
8GPT-4.1 minijudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.5
9Gemma 3 27B ITjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.4
10Gemini 2.0 Flash Litejudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8.2
11GPT-4o minijudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8
12Gemma 3 12B ITjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs8
13Gemini 1.5 Flashjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.9
14Qwen2.5 72B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.7
15GPT-4.1 nanojudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.6
16Gemma 2 27B ITjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.5
17Llama 4 Maverickjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.5
18Claude Haiku 3.5judge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.4
19Gemma 2 9B ITjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.3
20Qwen 2.5 14B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.1
21Gemma 3 4B ITjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs7.1
22Claude 3 Haikujudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs6.9
23Qwen2 72B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs6.7
24Llama 3 70B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs6.6
25Qwen2.5 7B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs6.5
26Llama 3.1 70B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs6
27Llama 4 Scoutjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs5.7
28Qwen2 7B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs5.6
29Llama 3.1 8B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs5.4
30GLM-4-9b-chatjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs5.3
31Qwen2.5 3B Instructjudge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs4.9
32Mistral (7B) Instruct v0.3judge:GPT-4o-2024-08-06· SeaLLMs Leaderboard for SEA Languages data docs2.7

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