CUDA Agent: AI Beats torch.compile and Claude Opus On KernelBench’s hardest Level‑3 tasks, the authors claim CUDA Agent beats torch.compile’s speed in ~92% of cases and outperforms Claude Opus 4.5 and Gemini 3 Pro by about 40 percentage points on the “faster than compile” rate. That’s not “LLM writes cute CUDA snippets.” That’s “an RL agent, with hardware in the loop, consistently out‑optimizes both compiler heuristics and frontier general models on real kernels.”

CUDA Agent outpaces torch.compile on 92% of hard kernels and beats Claude Opus & Gemini 3 Pro—RL + hardware is rewriting compiler rules. See the results #CUDAAgent #torchcompile #ClaudeOpus