AceGPT-v2-8B-Chat: benchmark results
FreedomIntelligence/KAUST's AceGPT v2 8B Chat, a dialogue-tuned Arabic-focused LLM fine-tuned from AceGPT-v2-8B, part of the AceGPT family spanning 7B to 70B. Provider: FreedomIntelligence. Released 2024-06-20. Access: Open.
Aggregate Rating 1514 ± 20, rank #729 of 1635 rated models, from 283 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Arabic LLM - Aratrust MentalHealth | 96.05 | Accuracy (%) | 97.2 |
| BALSAM - Program Execution | 93.78 | Overall score (0-100, LLM-judged generation and multiple cho | 91.9 |
| Open Arabic LLM - Alghafa Multiple Choice Facts Truefalse Balanced Task | 93.33 | Accuracy (%) | 86.7 |
| Open Arabic LLM - Arabic MMLU History (High School) | 49.21 | Accuracy (%) | 84 |
| BALSAM - Overall | 55.9 | Mean of category overall scores (0-100) | 83.3 |
| BALSAM - Logic | 38.08 | Overall score (0-100, LLM-judged generation and multiple cho | 80.6 |
| Open Arabic LLM - Arabic MMLU Economics (Middle School) | 79.31 | Accuracy (%) | 79.6 |
| BALSAM - Entailment | 78.83 | Overall score (0-100, LLM-judged generation and multiple cho | 77.4 |
| BALSAM - Text Manipulation | 62.26 | Overall score (0-100, LLM-judged generation and multiple cho | 77.4 |
| Open Arabic LLM - Alghafa Multiple Choice Rating Sentiment Task | 54.11 | Accuracy (%) | 75.3 |
| BALSAM - Fill in the Blank | 77.64 | Overall score (0-100, LLM-judged generation and multiple cho | 75 |
| Open Arabic LLM - Arabic MMLU Islamic Studies (Middle School) | 69.33 | Accuracy (%) | 74.1 |
Interactive version: theaggregate.ai/model?slug=acegpt-v2-8b-chat · How It Works · Data refreshed daily, snapshot 2026-10-15.