GPU Cloud Rental
Rent H100, H200, A100, RTX 4090, RTX 3090, RTX 3060, and RTX 2000 Ada cloud GPUs for AI training, inference, notebooks, ComfyUI, SDXL, FLUX, and vLLM workloads.
On-demand cloud GPUs with local pricing
Lumino GPU Cloud is built for AI builders who need short GPU runs without long contracts. Start a pod, connect with SSH, run your Docker image or notebook, and stop when the job finishes.
Popular GPU rental use cases
- Fine-tuning Llama, Qwen, Mistral, DeepSeek, and Gemma models
- Serving vLLM and TGI inference endpoints
- Running ComfyUI, Stable Diffusion, FLUX, and WAN video/image workflows
- PyTorch, JAX, TensorFlow, CUDA, and RAPIDS experiments
Transparent GPU rental pricing
Compare live GPU availability and hourly pricing before renting. Lumino supports per-second usage accounting, localized pricing, and no minimum monthly commitment.
Browse live GPU marketplaceGPU rental FAQ
Do I need a contract?
No. Add credits, rent a GPU when needed, and stop or terminate the pod after your workload is done.
Which GPUs are available?
Availability changes, but Lumino lists GPUs such as H100, H200, A100, RTX 4090, RTX 3090, RTX 3060, and RTX 2000 Ada when capacity is live.
Can I use my own Docker image?
Yes. You can bring a public Docker image and configure the pod for CUDA, notebooks, model serving, or training workflows.