Documentation ↗Excerpts from the project README on GitHub. Copyright and licensing remain with the respective authors.
GPUStack pools heterogeneous GPUs across machines into one cluster and schedules model workloads across them, with a web UI and OpenAI-compatible endpoints.
| Category | Run LLMs locally |
| Type | GPU cluster manager |
| License | Apache-2.0 |
| Runs locally | Yes |
| Built with | Python |
| Skill level | Advanced |
| Best for | teams with several GPU machines to pool |
Other open-source run llms locally tools worth comparing:
OllamaRun open LLMs locally from one command
JanOpen-source, offline ChatGPT-style desktop app
GPT4AllPrivate local AI that runs on CPU
llama.cppThe C/C++ engine powering local inference
LocalAIA drop-in OpenAI API you self-host
Text Generation WebUIFeature-rich web UI for local models
KoboldCppSingle-file local model runner
MLC LLMRun LLMs on any device, even phones
llamafileOne executable file = model + runtime
exoRun big models across your everyday devices
CortexOllama-style runtime from the Jan team
Nexa SDKRun any model on any device — CPU, GPU, NPU
RamaLamaRun models as OCI containersGPUStack is free and open-source (Apache-2.0 license), so you can use, self-host and modify it at no cost.
Yes. GPUStack is designed to run on your own machine or server, keeping your data private.
Popular open-source alternatives include Ollama, LM Studio, Jan. See the comparisons above to choose.
Browse the full directory of open-source AI tools, models and projects — updated daily.
Browse all tools →