Skip to main content

Welcome to Lium

Lium is agent-first compute: a GPU rental marketplace built so that an AI agent with only a shell can rent, run and release a GPU without a human in the loop. Providers contribute GPU resources to a global pool; renters access those resources on demand; Lium scores the GPU performance of every node to keep the pool honest.

Pick your audience​

Choose the section that matches what you're here to do:

AudienceI want to…Start here
ProvidersContribute GPU resources and get paidProviders →
RentersRent GPU compute for ML / data workloads on lium.ioRenters →
DevelopersBuild with the Lium API or CLIDevelopers →
AI agentsGo from no account to a running GPU pod, headlessAI Agents →

Compliance​

Lium enforces a strict ban on access to its compute from US embargoed countries and by sanctioned parties, and runs thorough checks to ensure its compute is never accessible from restricted areas. See the Export Control and Sanctions Policy.

Agent access​

This site is agent-native:

  • If you are an AI agent, start at AI Agents — the shortest path from no account to a running pod, on one page; then Scripting for agents for the flags, JSON and exit codes a script relies on.
  • Every page is reachable as <page-url>.md (e.g. /providers/quickstart.md).
  • Send Accept: text/markdown to any page URL to get raw markdown.
  • AI agents can use the /mcp endpoint (search + read_page tools).
  • /llms.txt and /llms-full.txt are at the site root per the llmstxt.org spec.

How Lium works​

Providers contribute GPU resources to a global pool. Users rent these resources for computational tasks — machine learning, data analysis, inference, training — and providers are paid based on the quality and performance of their GPUs.

  • Providers: Provide GPU resources, which Lium evaluates and scores.
  • Verification: Lium connects securely to provider machines to verify hardware specs and performance.
  • Renters: Rent computational resources through the lium.io platform.

Contact and support​