ParallelKernelBench: Frontier LLMs can't write fast multi-GPU kernels (yet)

Imported from official source

AI Classified by Officially

ParallelKernelBench tests whether LLMs can write fast multi-GPU CUDA kernels across 87 real workloads. The best model solves under a third, but a few generated kernels beat any public implementation.

This is an extract. The publication continues at the source.

Read the original at the source: https://www.together.ai/blog/parallelkernelbench

Officially imported this from Together AI’s own source and shows an extract. If you work there, claiming the profile and verifying the domain lets you choose to show the full text here.

Provenance

Organization
Together AI — imported from official source
Official source
https://www.together.ai/blog/rss.xml RSS
Imported
September 20, 2026 19:52
Versions
1 recorded
Identity
https://www.together.ai/blog/parallelkernelbench

Officially records where a publication came from, not whether it is true. Imported records are reproduced from an organization's own official source.