Mistral Small

Mistral Small, a 22-billion parameter model from mistralai, offers a compelling mid-point solution with a 32768 context length. Launched on January 9, 2024, it performs among the fastest models, typically ranking in the top tier for speed (77th percentile). It also offers competitive pricing, generally providing cost-effective solutions (69th percentile). Demonstrating exceptional reliability, Mistral Small boasts a 99% success rate across benchmarks, indicating minimal technical failures. In terms of specific performance, Mistral Small excels in hallucination prevention, achieving perfect 100% accuracy, making it the most accurate model at its price point and speed. It also shows strong performance in Email Classification (96.0% accuracy). However, its performance in more complex cognitive tasks is moderate to low. General Knowledge (91.5% accuracy) and Ethics (86.0% accuracy) are in the lower percentiles. Instruction Following (52.0% accuracy), Reasoning (30.0% accuracy), and Coding (41.0% accuracy) represent notable weaknesses, with scores placing it in the lower percentiles for these categories. While it offers multilingual support and improved reasoning capabilities, its benchmark results suggest these are areas for further development.