auto-gpu-kernel 1.0 generates GPU kernels and harnesses
Version 1.0 introduces an autonomous meta-harness that evolves both GPU kernels and the harness layer; the project won the DSA category in an MLSys 2026 FlashInfer contest.
The author announced version 1.0 of auto-gpu-kernel, described as an autonomous meta-harness for kernel generation. According to the project page, it evolves both the kernel and the harness layer. The project won the DSA category in the MLSys 2026 FlashInfer agent-only kernel-generation contest.
For engineers, the result is a starting point for evaluating tools that generate GPU kernels and harnesses. The contest participation and win are facts about the project, not a guarantee of performance on other workloads. If you use AI to study or apply this material, avoid sending proprietary code, customer data, credentials, or other personal data; use synthetic or de-identified examples and review results before running them.