Run anything in the cloud

vmux run bundles your code and runs it in a long-running container. attach from anywhere, like tmux. (free for the holidays, then $2/mo)

your laptop →

🐰 bundler → cf container → r2

your laptop ↓

🐰 bundler ↓ cf container ↓ r2

↳ tmux session (attach from anywhere)

uv tool install vmux-cli

vmux run python train.py
# like uv run, but in the cloud. you get a tmux session (attach from anywhere).

vmux run -d python train.py
# detached. close your laptop. job keeps running (up to 7 days for now).

vmux attach abc123
# back in your tmux session. from your phone, another laptop, wherever.

vmux run -d -p 8000 python server.py
# expose a port, get a preview URL. websockets just work.

vmux ps
vmux logs -f abc123
vmux stop abc123

how it works

AST parses imports, bundles only what you need
auto-detects deps from pyproject.toml, requirements.txt, or PEP 723 inline scripts
editable packages? auto-detected and bundled with transitive deps
runs on cloudflare containers (indefinitely soon)
pre-baked: pytorch, transformers, numpy, pandas, fastapi

isolated containers · full network · websockets · real tmux

no dockerfile · no server · no ssh · gpus soon

performance

cold start: ~4s
detached start: ~500ms
bundler: 353MB → 42KB (skips .git, node_modules, venvs)

limits

0.25 vCPU, 1GB RAM per job
2GB disk (ephemeral, R2 mount soon)
7 day max runtime
outbound unrestricted, inbound via preview URLs
experimental: jobs may be randomly evicted

faq

what does beta mean? free to use, APIs may change, jobs may be evicted. use for dev, not prod yet.
who is behind this? surya. ml engineer, previously built sub-50ms vector databases.
what's the roadmap? 1. stability (#317, #319) 2. 🐰