8.8
/ 10
DeepSeek R1
429 USER VOICES · REALITY SCORE
A high-performance reasoning model excelling in math and coding via pure RL, though local inference is slow and censorship filters vary by d