Audience
ML engineers, AI researchers and startups renting GPUs by the hour
About Lium
Lium is a marketplace for renting NVIDIA GPUs by the hour. Providers list their machines and compete on price, so the rates come from the market rather than from us.
You can rent a single GPU or a whole 8-GPU node and have a container running in about a minute, with SSH and Jupyter access. Today's floors are $0.16/hr for an RTX 3090, $0.27 for a 4090, $0.49 for a 5090 and $4.23 for an H200. Billing is per second, so a 12-minute job costs 12 minutes.
There is a CLI and a Python SDK for scripting deployments, and a free public API with live prices and availability. New accounts start with $5 of credit, and nothing here requires a contract or a minimum commitment.
Built for ML training, fine-tuning, inference and batch jobs where hyperscaler pricing makes no sense.
Other Popular Alternatives & Related Software
AIC Cloud
AIC Cloud is a global cloud hosting platform serving developers, startups, and small businesses across Asia, Africa, Europe, and North America. Linux VPS from $1.19/month on NVMe SSD with full root access, bare-metal dedicated servers from $22/month, NVIDIA cloud GPU instances (RTX 3090, RTX 4090, A100) billed per minute from $0.14/hour, business email with custom domains, and one-click app deployment.
What separates AIC Cloud: flat monthly pricing - the rate you sign up at is the rate you pay every month, per-minute GPU billing with no hourly minimums, real-human WhatsApp and email support, multi-region datacenters across Asia Pacific, Europe, and North America with anti-DDoS protection on every plan, and multiple payment methods including UPI, international cards, and net banking.
Built by Applied Intelligence Corporation, a bootstrapped technology company - operating since 2024.
Learn more
NVIDIA Brev
NVIDIA Brev is a cloud-based platform that provides instant access to fully configured GPU environments optimized for AI and machine learning development. Its Launchables feature offers prebuilt, customizable compute setups that let developers start projects quickly without complex setup or configuration. Users can create Launchables by specifying GPU resources, Docker images, and project files, then share them easily with collaborators. The platform also offers prebuilt Launchables featuring the latest AI frameworks, microservices, and NVIDIA Blueprints to jumpstart development. NVIDIA Brev provides a seamless GPU sandbox with support for CUDA, Python, and Jupyter Lab accessible via browser or CLI. This enables developers to fine-tune, train, and deploy AI models with minimal friction and maximum flexibility.
Learn more
Packet.ai
Packet.ai is a GPU cloud platform built to give developers and AI teams fast access to high-performance computing without the complexity and inefficiencies of traditional cloud infrastructure. It provides on-demand GPU instances, including modern NVIDIA hardware, that can be launched in seconds and accessed through tools like SSH, Jupyter, or VS Code, enabling users to quickly start training models, running inference, or experimenting with AI workloads. It introduces a different approach to GPU usage by dynamically allocating resources based on real-time workload demands, rather than treating a GPU as a fixed unit, allowing multiple compatible workloads to share hardware efficiently while maintaining predictable performance. This results in higher utilization and eliminates the need to pay for idle capacity, focusing instead on the exact compute resources consumed. Packet.ai also offers an OpenAI-compatible API for language model inference, embeddings, and fine-tuning, etc.
Learn more
Trooper.AI
Trooper.AI lets you rent private, bare-metal GPU servers for AI training, inference, and experimentation — ready in minutes. Instantly deploy OpenWebUI, ComfyUI, Jupyter Notebook, Ubuntu Desktop, Ollama, and more with one click. No shared GPUs, no containers, full root access included.
All servers are EU-hosted, GDPR and EU AI Act compliant, and operated from Germany. Trooper.AI is built on up-cycled high-end hardware, combining strong performance with sustainability.
Pause or freeze servers anytime to save costs and pay only for what you use. Choose from a wide range of GPUs, from V100 and RTX 3090 to RTX 4090 and RTX Pro 6000 Blackwell, backed by fast NVMe storage, persistent machine state, automatic backups, and simple UI and API management.
Trooper.AI is the smallest hyperscaler in Europe — built for developers who want performance, privacy, and full control without cloud complexity.
Learn more
Pricing
Starting Price:
$0.16 per GPU per hour
Integrations
No integrations listed.
Company Information
Lium
lium.io
Videos and Screen Captures
Other Useful Business Software
$300 Free Credits to Build on Google Cloud
Put your $300 in credit toward real workloads, then keep building with free monthly usage for 20+ products. No commitment and no charge until you upgrade.
Product Details
Platforms Supported
Cloud
Training
Documentation
Support
Online