Mistral Medium

Mistral Medium, a closed-source model from Mistral AI, demonstrates strong overall performance, particularly excelling in reliability and speed. Created on January 9, 2024, with a substantial 32768 context length, it consistently ranks among the fastest models, achieving the 86th percentile in speed across benchmarks. This is further evidenced by its 92534ms duration in the Reasoning (Baseline) benchmark, placing it in the 86th percentile for speed within that specific test. The model also exhibits exceptional reliability, scoring in the 100th percentile, indicating minimal technical failures and a consistent ability to provide usable responses. In terms of cost, Mistral Medium offers moderate pricing, ranking in the 24th percentile overall and for the Reasoning (Baseline) benchmark ($0.0565). While its 73.0% accuracy in the Reasoning (Baseline) benchmark places it in the 71st percentile, indicating solid performance in complex reasoning tasks, there is room for improvement to reach top-tier accuracy. Its key strengths lie in its remarkable reliability, impressive speed, and strong reasoning capabilities, making it a robust choice for applications requiring consistent and quick responses.