CoreWeave

Open source activity

As of Oct 28, 2026

CoreWeave's recent repos read exactly like what you'd expect from a GPU-cloud provider: infrastructure-as-code (a Terraform provider for its own platform), distributed-training tooling (NCCL tests for multi-GPU communication), and ML-ops plumbing — container images and Slurm-on-Kubernetes containers for running batch training jobs.

Two very recently touched repos point somewhere newer: serverless-inference-for-hermes and serverless-inference-openclaw-plugin suggest CoreWeave is building a serverless inference layer on top of its raw GPU rental business, rather than just renting bare-metal capacity by the hour. That would put it in more direct competition with managed-inference offerings from the hyperscalers, not just with other neocloud GPU lessors.

Most active repositories