Create high-performance GPU kernels for state-of-the-art LLM architectures on NVIDIA Blackwell GPUs with humans and/or AI agents
Create AI agents that automatically generate optimized CUDA kernels for cutting-edge LLM operations. Your agent will receive kernel specifications and must produce high-performance code for NVIDIA Blackwell B200 GPUs.
Compete across workloads derived from production models. Kernels are evaluated on correctness, speed, and win rate against FlashInfer baselines.
We welcome both expert-crafted seed kernels with agent-assisted evolution, and fully agent-generated solutions. The two approaches will be evaluated separately. Agent solutions must open-source scripts to reproduce kernels. No API credits provided.
Three kernel categories targeting the most important operations in modern LLMs
Fused Mixture-of-Experts kernels with FP8 support.
Everything you need to start competing
Agent interface specifications and kernel solution requirements.
Coming SoonScoring metrics, correctness thresholds, and ranking methodology.
Coming SoonFlashInfer production kernels and OpenEvolve-based references.
Coming SoonGPU cards for top performing teams. Details coming soon.
Winners receive complimentary MLSys 2026 conference registration.
Registered teams receive Modal compute credits for NVIDIA B200 GPU development.
Join teams from around the world in pushing the boundaries of AI kernel generation.
Register Your TeamRegistration deadline: February 15, 2026
