Venice: Uncensored

Venice: Uncensored, a fine-tuned variant of Mistral-Small-24B-Instruct-2501, demonstrates a strong overall performance profile. It consistently ranks among the fastest models, placing in the 85th percentile across all benchmarks, and offers competitive pricing, typically falling within the 67th percentile. A standout feature is its exceptional reliability, achieving a 100% success rate across all evaluated benchmarks, indicating minimal technical failures. In terms of specific capabilities, the model excels in General Knowledge and Ethics, both achieving 99.0% accuracy, though their percentile rankings suggest a competitive landscape. Email Classification also shows strong performance at 99.0% accuracy (69th percentile). However, the model exhibits notable weaknesses in Reasoning (57.1% accuracy, 32nd percentile) and Mathematics (81.0% accuracy, 33rd percentile), suggesting areas for improvement in complex problem-solving. Its ability to acknowledge uncertainty in the Hallucinations benchmark is moderate at 89.8% accuracy (34th percentile). Instruction Following and Coding show moderate performance, with accuracies of 59.1% and 81.8% respectively. Overall, Venice: Uncensored is a highly reliable and fast model with strong general knowledge and ethical reasoning, but could benefit from enhancements in complex reasoning and mathematical capabilities.