Sandboxes / every shape
Shared boxes for burst work, dedicated boxes with no neighbours, environments built for RL rollouts, browser sandboxes that reach real sites, and an execution API with the heavy Python stack already installed. Need something we don't list — we'll build it with you.
no sandbox — create one
< 5 ms
step latency · custom RL
~1 s
to a running machine
isolated
from your infrastructure
per-second*
after the first hour
Live instances
4 activeThe catalogue
Pooled capacity for short work. Starts in about a second, with a one-hour minimum and per-second billing after that. The default for agent tool-calls and one-off runs.
Pinned vCPU, memory and disk on a host that is yours alone. Predictable latency for workloads that hate noisy neighbours — long agent sessions, benchmarks, anything you time.
Thousands of identical environments kept warm between steps and stepped in one batch. Soft reset is in-process; hard reset respawns clean. Fast steps without the memory cost of holding snapshots.
Your simulator, your image, your hardware profile — co-designed with us and pinned to reserved capacity. Sub-5 ms steps, custom observation plumbing, whatever the training loop needs.
A disposable browser that reaches the real web, so an agent can open ChatGPT, Perplexity, Gemini or any site, act on it, and hand you back the session. Nothing touches your own machine.
Send code, get the result back. Judge0-style execution with the heavy Python stack — PyTorch, NumPy, SciPy, pandas — already resident, so nothing waits on an install.
Command line
The installer detects your OS and CPU, verifies a checksum and drops the bzlabs binary on your PATH. macOS and Linux, Intel and ARM;
Windows works under WSL.
curl -fsSL https://boltzlabs.john221wick.in/cli/install.sh | shTell us the shape