Magnum v4 72B

Magnum v4 72B, developed by anthracite-org and fine-tuned on Qwen2.5 72B, demonstrates moderate speed performance, ranking in the 32nd percentile across benchmarks. Its pricing is also moderate, placing it in the 24th percentile. A standout feature is its exceptional reliability, achieving a 100% success rate across all evaluated benchmarks, indicating consistent and stable operation. The model exhibits remarkable strength in specific areas. It achieved perfect accuracy in both Hallucinations (Baseline) and Ethics (Baseline) tests, effectively avoiding fabricated information and consistently identifying ethical responses. For these categories, it is noted as the most accurate model at its price point and among models of comparable speed. In General Knowledge, it scored a respectable 98.5% accuracy. Its performance in Email Classification was solid at 97.0%, while Reasoning presented a moderate 68.0% accuracy. Overall, Magnum v4 72B excels in tasks requiring precision, ethical judgment, and the ability to acknowledge uncertainty, making it a strong contender for applications where factual integrity and responsible AI behavior are paramount.