DeepSeek and OpenAI’s o1 models performed the best across the various benchmarks, but all models still struggle in a range of ...