deepseek repeating itself

deepseek reinforcement

how does deepseek v3 compare to other large language models in terms of performance