AgentKernelArena: Generalization-Aware Benchmarking of GPU Kernel Optimization Agents

In three linesAgentKernelArena is an open-source benchmark for evaluating AI coding agents on GPU kernel optimization. It contains 196 tasks (HIP-to-HIP, Triton-to-Triton, PyTorch-to-HIP) and tests generalization to unseen configurations. Cursor Agent, Claude Code, and Codex Agent achieve speedups up to 6.89x, but PyTorch-to-HIP optimizations show correctness drops on unseen configurations.

Summary generated by Claude — human-verified