Give your agent GPUs.
Givemeanode is a slurm-like GPU orchestrator built for AI agents like claude, codex, or opencode. Queue up thousands of jobs on large scale GPU clusters, get billed by the minute. Read the docs.
claude mcp add --transport http givemeanode https://mcp.givemeanode.comUsing Claude in the browser? Go to Settings, Connectors, Add custom connector and enter the remote MCP server URL https://mcp.givemeanode.com.
codex mcp add givemeanode --url https://mcp.givemeanode.comIn ~/.cursor/mcp.json:
{
"mcpServers": {
"givemeanode": { "url": "https://mcp.givemeanode.com" }
}
}In opencode.json (project or global):
{
"mcp": {
"givemeanode": {
"type": "remote",
"url": "https://mcp.givemeanode.com",
"enabled": true
}
}
}https://mcp.givemeanode.comdevin mcp add givemeanode https://mcp.givemeanode.comOr in ~/.config/devin/config.json:
{
"mcpServers": {
"givemeanode": { "url": "https://mcp.givemeanode.com", "transport": "http" }
}
}In Capy, go to Settings, MCP servers, New MCP server, pick the HTTP transport, and enter the server URL:
https://mcp.givemeanode.comThen open the saved server and click Connect to finish the OAuth sign-in.
In ~/.config/amp/settings.json (or the Amp extension settings in VS Code):
{
"amp.mcpServers": {
"givemeanode": { "url": "https://mcp.givemeanode.com" }
}
}In ~/.omp/agent/mcp.json (or a project's .omp/mcp.json) — /mcp add in a session does this for you:
{
"mcpServers": {
"givemeanode": { "type": "http", "url": "https://mcp.givemeanode.com" }
}
}In Zed's settings.json (zed: open settings from the command palette):
{
"context_servers": {
"givemeanode": { "url": "https://mcp.givemeanode.com" }
}
}Zed prompts for the OAuth sign-in on first use.
What people use it for
Each of these is one sentence to your agent. The docs show every example as English, CLI, and the MCP calls underneath.
Run anything on an H100 in one sentence
“Get me a GPU node, clone github.com/you/yourrepo on it, and run the benchmark.”
A container on an H100 with a persistent disk, billed by the minute. See nodes.
Fine-tune a model overnight
“Start the fine-tune on the node in the background, check the loss every hour, and stop the node when it converges.”
Training runs detached and outlives the conversation. Stopping the node stops the bill with the files intact, like tmux on a machine you can power off. See long-running commands.
Sweep and evaluate as batch jobs
“Sweep learning rates 1e-4, 3e-4, and 1e-3 over the training job as a batch, and show me a results table when they finish.”
Up to 256 variants in one call, billed only while running, with queued time free. Jobs can checkpoint, so a preempted run resumes where it left off and you pay only for progress that survives. Each job declares its result as JSON, so a finished sweep reads as a table. See batch jobs.
Serve a model at a public URL
“Serve the merged model with vLLM and give me a URL the eval harness can call.”
Any port on a running node gets a public HTTPS URL, with optional bearer auth. Traffic is free. See exposing ports.
Pricing
| Item | Rate | Minimum / free tier | Discount |
|---|---|---|---|
| 1×H100 node | $0.06/min ($3.60/hr) | - | - |
| 8×H100 node | $0.48/min ($28.80/hr) | - | - |
| Batch jobs (H100) | $0.045/min per GPU | - | 25% below interactive |
| Job builds | $0.01/min | 300 min/month free | - |
| Logs | $0.50/GB ingested | 5 GB/month free | - |
| Metrics | $8 per 1,000 active series/month | first 10,000 series free | - |
| Traces | $0.50/GB ingested | 5 GB/month free | - |
| Snapshot storage | $0.10/GiB-month | 250 GiB free | - |
| Object storage | $0.10/GB-month | 100 GB free | - |