Instant cloud sandboxes,
built for AI engineers
BlueMachine is built to empower AI developers, researchers, and builders to execute python scripts, run heavy model fine-tuning, or evaluate untrusted code inside isolated, secure sandboxes. Zero configuration, absolute control.
An intelligent stack for
autonomous operations
BlueMachine is built to empower AI developers, researchers, and engineers to execute complex compute decisions, orchestrate remote actions, or automate pipelines inside highly reliable, secure sandboxes. Zero complexity, absolute reliability.
Terminal client + high-speed sync
Run command sequences in isolated cloud workspaces directly from your terminal. Our client generates ephemeral SSH keys and initiates parallelized rsync compression to transfer local files in less than a second.
Only modified file deltas are transferred, ensuring zero delay. When the remote script finishes execution or reaches its maximum lease duration, the VM is safely destroyed automatically.
Secure, isolated compute environments
Run code on dedicated, hardware-isolated micro VM instances on Google Cloud Platform. Each sandbox gets root access, custom Docker images, and dedicated networking routes with no risk of container escapes.
Enforce exact runtime limits using per-second billing with auto-teardown. Custom TTL limits and idle network checks prevent wasteful compute costs and ensure absolute containment.
GPU-accelerated VM sandboxes
Run heavy model training, fine-tune neural nets, or execute large-scale LLM inference inside isolated VM nodes equipped with dedicated Nvidia GPUs (T4, A100). Provisioned in less than 5 seconds.
Take advantage of precise per-second billing on heavy workloads. Sandboxes automatically shut down once active execution finishes, protecting you from costly idle GPU time.
Choose the perfect execution container. Scale instantly from single-core CPU micro-VMs up to massive Nvidia A100 GPU clusters.
Spark Nano
0.2 vCPU (shared) • 0.6 GB RAM
Minimal container for basic monitoring agents, lightweight cron jobs, and text scrapers.
Spark Entry
0.5 vCPU (shared) • 1.7 GB RAM
Great for quick webhook handlers, API pings, and basic background tasks.
Spark Micro
0.25 vCPU (burst to 2) • 1 GB RAM
Perfect for lightweight scripts, cron automation, API health checks, and small scrapers.
Spark Small
0.5 vCPU (burst to 2) • 2 GB RAM
Ideal for basic API testing, small python tasks, and low-memory background functions.
Spark Medium
1 vCPU (shared) • 4 GB RAM
Burst-capable shared CPU instance, perfect for slightly heavier periodic scripting tasks.
Spark Dedicated
1 dedicated AMD vCPU • 4 GB RAM
Dedicated AMD Milan core for predictable runtime performance and zero CPU contention.
Estimate your serverless sandbox compute budget in real-time. Billed down to the exact second of active execution.
This is an illustrative example to help you estimate your compute costs; actual amounts may vary depending on execution environments, network bandwidth, and storage variables.
Start running in minutes
Integrate via CLI or orchestrate VM lifecycles using Node.js & Python SDKs.
# 1. Install global client utilitynpm install -g @bluemachine/machine# 2. Login to your accountmachine login# 3. Lease VM, sync folder, and run immediatelymachine run "python train.py --epochs 3"
Engineered for speed, built for containment
Bluemachine delivers hardware-virtualized sandboxes, block-level directory synchronization, and second-precision billing down to the millisecond. Check out our real-time active system telemetry inside each drawer.
Zero waste. Absolute budget efficiency.
Only pay down to the exact millisecond that your sandboxed VM lease is active. The moment your remote commands finish executing, or your idle timeout thresholds are crossed, the cloud hardware resources are immediately reclaimed. No monthly retainers, no idle compute waste.
Near-instant code delta synchronization.
Integrate programmatically and sync your workspace code in under a second. Our lightweight developer client performs block-level file comparisons over highly optimized parallel SSH links. Only modified delta changes are transferred, achieving directory synchronization speeds under one second.
Dedicated security boundaries.
Ditch insecure container boundaries and shared kernels. Bluemachine sandboxes run on dedicated, hardware-isolated micro VM instances on Google Cloud Platform. Get root access, custom Docker images, and dedicated networking routes with zero risk of container escapes.
Developer APIs for AI agent workflows.
Easily spin up, monitor, write files to, and tear down secure sandbox environments dynamically from within your python systems or javascript agent networks using our unified Developer APIs and clean SDK clients. Web node latency fluctuates in real-time, routing cursors drift continuously.
Flexible compute packages designed to support creators, engineers, and businesses at different stages of their scale
Free Plan
Target audience: Testing & Hobby+ Per-second Spark VM compute lease billing.
- 5 sandbox runs / day
- Max 10 mins session TTL
- All Spark VMs access
- High-speed directory sync
- Developer SDK & API access
Starter Plan
Target audience: Small Projects+ Per-second Spark & Flex VM compute lease billing.
- 50 sandbox runs / day
- Max 60 mins session TTL
- All Spark & Flex VMs access
- Developer SDK & API access
- Parallel run pipeline
Pro Plan
Target audience: Production Scaling+ Per-second Spark, Flex & Prime VM compute lease billing.
- Unlimited sandbox runs / day
- Max 240 mins (4 hr) TTL
- All Spark, Flex & Prime VMs access
- Developer SDK & API access
- High priority run queues
GPU Plan
Target audience: Heavy ML Tasks+ Per-second compute VM billing (includes Dedicated Nvidia GPUs).
- Unlimited sandbox runs / day
- Max 120 mins session TTL
- All CPU & Dedicated Nvidia GPU VMs
- Dedicated GPU queues
- SLA priority support
Not sure which plan fits you?
Talk to our cloud operations engineers. We can customize session timeouts, provision dedicated CPU/GPU pools, establish custom SLA metrics, and offer bulk pricing for large-scale enterprise deployments.
Still have questions?
If you didn't find the answers you were looking for, please contact our support engineers or consult our developer documentation guides. We're here to help you deploy sandboxes.
Ready to accelerate your workflows?
Launch secure, isolated micro VM sandboxes programmatically or sync your terminal workspace in under 3 seconds. Per-second billing down to the millisecond. Zero waste.