GoTo-AI Leaderboard - IndoMMLU - STEM: leaderboard
GoTo-AI Leaderboard (GoToCompany, successor of the Sahabat-AI leaderboard), evaluating mostly open 7-72B models. IndoMMLU (Koto et al., 2023) questions on STEM subjects, score on 0-100 (higher is better); published for instruct and reasoning models only.
Metric: Score (0-100). Source: huggingface.co. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek R1 Distill Llama 70B | 76.84 | #614 |
| 2 | DeepSeek R1 Distill Qwen 32B | 76.6 | #662 |
| 3 | Qwen 2.5 72B Instruct | 63.62 | #469 |
| 4 | Llama-SEA-LION-v3-70B-IT | 62.27 | #407 |
| 5 | Llama 3.1 70B Instruct | 61.24 | #567 |
| 6 | Llama 3.3 70B Instruct | 60.66 | #541 |
| 7 | Gemma 3 27B (IT) | 59.89 | #528 |
| 8 | gemma2-9B-cpt-sahabatai-v1-instruct | 53.84 | #618 |
| 9 | Gemma 2 9B (IT) | 51.07 | #795 |
| 10 | Qwen 2.5 7B Instruct | 45.48 | #863 |
| 11 | Llama 3.1 8B Instruct | 39.19 | #1017 |
| 12 | Llama 3 8B Cpt Sahabatai V1 Instruct | 33.23 | #905 |
| 13 | Llama 3 8B Instruct | 30.75 | #1116 |
| 14 | Qwen 2 7B Instruct | 19.09 | #982 |
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-indommlu-stem · How It Works · Data refreshed daily, snapshot 2026-11-05.