Arcee AI: Virtuoso Large

Arcee AI's Virtuoso-Large is a 72B parameter general-purpose LLM designed for cross-domain reasoning, creative writing, and enterprise QA, leveraging a 128k context length inherited from Qwen 2.5. It consistently ranks among the fastest models, performing in the 92nd percentile across 8 benchmarks, and offers competitive pricing, placing in the 44th percentile. The model demonstrates exceptional reliability with a 99% success rate, indicating minimal technical failures. Virtuoso-Large exhibits outstanding performance in Hallucinations, Ethics, and Email Classification, achieving perfect accuracy in the former two and 99% in the latter, often being the most accurate at its price point or speed. It also shows strong capabilities in General Knowledge (99.5% accuracy) and Coding (84% accuracy). While its Instruction Following (63% accuracy) and Reasoning (74% accuracy) scores are respectable, they are not its strongest suits. Mathematics (79% accuracy) falls in the middle range. Its aggressive KV-cache optimizations contribute to low-second first-token latency on H100 nodes, making it a practical production-grade solution, particularly as a fallback in Conductor pipelines.