Timeline of deepseek-v3

deepseek-v3 scores 16.85% on Terminal-Bench

deepseek-v3 scores 0% on Terminal-Bench

deepseek-v3 scores 22.81% on τ²-bench

deepseek-v3 scores 34.76% on IFBench

deepseek-v3 scores 35.76% on SciCode

deepseek-v3 scores 2.9% on Humanity's Last Exam

deepseek-v3 scores 55.66% on GPQA

deepseek-v3 scores 8.49 on Artificial Analysis Intelligence Index