Sahabat-AI - Indonesian Average: leaderboard
Sahabat-AI multilingual Indonesian benchmark overall — average across its Indonesian, Javanese and Sundanese evaluation tracks.
Metric: Average Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | gemma2-9B-cpt-sahabatai-v1-instruct | 59.91 | #608 |
| 2 | Llama 3 8B Cpt Sahabatai V1 Instruct | 49.98 | #910 |
| 3 | Gemma 2 9B (IT) | 48.85 | #788 |
| 4 | Gemma 2 9B | 43.35 | #958 |
| 5 | Llama 3.1 8B | 39.8 | #1151 |
| 6 | Llama 3 8B | 38.52 | #1242 |
| 7 | Llama 3 8B Cpt Sea Lionv2 Base | 38.5 | #1182 |
| 8 | Llama 3 8B Cpt Sea Lionv2.1 Instruct | 38.29 | #1024 |
| 9 | Llama 3.1 8B Instruct | 37.74 | #1035 |
| 10 | Qwen 2.5 7B Instruct | 37.74 | #862 |
| 11 | Qwen 2.5 7B | 37.53 | #922 |
| 12 | Qwen 2 7B | 33.26 | #1095 |
| 13 | Qwen 2 7B Instruct | 30.58 | #991 |
| 14 | Llama 3 8B Instruct | 30.44 | #1129 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=sahabat-ai-indonesian-average · How It Works · Data refreshed daily, snapshot 2026-10-20.