Skip to content
runcrate
Console

Research and experiments

Try an idea. Keep it or shut it down.

Open Jupyter on an H100, run the ablation, save results to a volume and delete the instance. Most GPUs bill by the second; a few by the minute.

Launch an instanceSee GPU prices
L40Sfrom $1.26/hr

What you can build

Three jobs people run here. Each on a product you can start today.

Crate

Jupyter and VS Code ready

The Devbox crate is a ready image your instance boots, with Jupyter, VS Code Server, Python with uv and Docker.

See what an instance includes
Instance

Parallel experiments

Launch one instance per configuration for an ablation or sweep, compare the runs, and delete the ones that lose.

Compare GPUs
Storage

Results that outlive the instance

Save notebooks, weights and logs to a volume, delete the instance, and mount the volume again tomorrow.

Read how volumes work

Examples

Ask your agent. Or read the code it writes.

Connected by the line, Claude Code or Codex calls the Runcrate tools for you: launch, copy files, run, delete.

Connect your agent first

Add Runcrate's MCP server to yourself: run curl -fsSL"https://runcrate.ai/mcp.md"and follow it.

Add to CursorInstall in VS Code
Works inClaude CodeCodexOpenCodeClineRoo CodeKilo CodeCursor

Docs
Start an A100 with the Devbox crate, run repro.sh to reproduce the paper's results, then delete the instance. Calls the Runcrate tool create_instance. It returns: status deploying: 1x A100. Calls the Runcrate tool instance_status. It returns: status running, with an address. Calls the Runcrate tool file_upload. It returns: repro.sh, on the instance. Calls the Runcrate tool ssh_execute. It returns: exit code 0: the paper's results, reproduced. Calls the Runcrate tool delete_instance. It returns: deleted: true.

Models and tools

Research tools. All of them run on your instance.

NameRoleRuns on
JupyterNotebooks in the browserInstance
VS Code ServerEditor in the browserInstance
PyTorchModels and trainingInstance
Weights & BiasesTrack every runInstance
OptunaHyperparameter searchInstance
Ray TuneSweeps across instancesInstance

How it works

From first call to production. Three steps.

  1. 01

    Start an environment

    Launch an instance with the Devbox crate and open Jupyter or VS Code in your browser. Clone your repo and install what it needs.

  2. 02

    Run the experiment

    Train, ablate or sweep, and watch metrics as they come in. Launch parallel instances when you need to compare configurations.

  3. 03

    Save results, shut down

    Write notebooks, weights and logs to a volume, then delete the instance. Start a new one when the next idea arrives.

Pricing

What it costs. Live prices, as billed.

See every GPU price

Name

As billed

  • L40S

    GPU, 48GB, cheapest offer

    $1.26/hr

  • A100 80GB

    GPU, 80GB, cheapest offer

    $1.43/hr

  • H100

    GPU, 80GB, cheapest offer

    $2.85/hr