GoTo-AI Leaderboard - Balinese: leaderboard
GoTo-AI Leaderboard (GoToCompany, successor of the Sahabat-AI leaderboard), evaluating mostly open 7-72B models. Balinese language score from its SEA-HELM-style suite: mean of competency scores (sentiment NLU, Indonesian-Balinese translation both ways, instruction following), with classification tasks normalized against the random baseline to 0-100; higher is better.
Metric: Score (0-100). Source: huggingface.co. Saturation forecast: Estimated already saturated. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 3 27B (IT) | 64.05 | #528 |
| 2 | Llama-SEA-LION-v3-70B-IT | 58.64 | #407 |
| 3 | Llama 3.3 70B Instruct | 54.31 | #541 |
| 4 | Llama 3.1 70B Instruct | 53.65 | #567 |
| 5 | DeepSeek R1 Distill Llama 70B | 50.66 | #614 |
| 6 | Gemma 2 9B (IT) | 47.95 | #795 |
| 7 | Llama 3.1 8B | 45.68 | #1130 |
| 8 | Gemma 2 9B | 43.98 | #924 |
| 9 | Qwen 2.5 72B Instruct | 43.07 | #469 |
| 10 | gemma2-9B-cpt-sahabatai-v1-instruct | 42.2 | #618 |
| 11 | Llama 3 8B | 42.06 | #1228 |
| 12 | Qwen 2.5 7B | 40.06 | #895 |
| 13 | Qwen 2 7B | 39.25 | #1070 |
| 14 | Llama 3 8B Cpt Sahabatai V1 Instruct | 38.8 | #905 |
| 15 | Llama 3.1 8B Instruct | 35.96 | #1017 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=goto-ai-leaderboard-balinese · How It Works · Data refreshed daily, snapshot 2026-11-05.