UE5 simulation, ML/RL training, or any Docker-based GPU workload.
Distributed across consumer GPUs. Results in minutes, not hours.
Upload your project (.zip or .7z) through our web client. Resumable uploads handle files up to 50 GB.
Choose your mode: Annotations, UE5+Python Training, Parameter Sweeps, or Docker Only. Configure scripts, hyperparameters, and outputs.
Our distributed GPU network runs your training in secure Docker containers with full GPU passthrough.
Get your trained .onnx model, training logs, and ML-ready datasets. All data is wiped from compute nodes after delivery.
Provision, configure, manage infrastructure. Pay even when idle.
Maintenance, power, cooling, depreciation. Years to ROI.
Distributed consumer GPUs. Flat $0.75/hr with hardened Docker containers.
Submit a parameter grid, run N variants across N GPUs simultaneously
Run any GPU workload without UE5 — IsaacSim, Gazebo, PyBullet, custom ML training
Upload YOUR UE5 project, not ours. Your IP stays yours.
Consumer GPUs working together. RTX 3060 to RTX 5090.
RGB, Depth, Segmentation, 2D/3D BBox, LiDAR, Tracking IDs — 11 types total
PPO, SAC, TD3 via Stable Baselines3. Train agents in your UE5 environment.
Hardened Docker containers with capability dropping and resource isolation
Every compute node has a unique cryptographic identity. Hardware fingerprinting prevents spoofing.
Cryptographic verification that every frame was actually computed. Operators stake reputation on honest execution.
All project data, intermediate files, and render output are securely wiped from operator nodes after every job completes.
All compute nodes verified as US-based via IP geolocation and Cloudflare. Your data never leaves American soil.
Every uploaded project scanned for malware and zip bombs before dispatch to compute nodes
Pay only for actual GPU-seconds consumed. No rounding up, no minimum time blocks. $0.50 minimum charge.
Run any GPU workload without UE5: IsaacSim, Gazebo, PyBullet, custom ML training — just Docker
| Rate | What You Get |
|---|---|
| $0.75/GPU-hour | Hardened Docker containers, GPU passthrough, ClamAV scanning, Proof of Compute, secure wipe after delivery |
Credits are stored as dollars at $0.75/GPU-hour.
| Credit Pack | GPU Hours |
|---|---|
| $25 | ~33 hours |
| $100 | ~133 hours |
| $250 | ~333 hours |
Free trial: $5.00 credit (~6 GPU-hours), no credit card required
Per-second billing. 20% platform fee — operators receive 80%. All compute US-only.
Start Training — Free $5 CreditInstall SimulacrumNode, connect your GPU, earn money while you sleep. Supports RTX 3060 through RTX 5090. Seamless auto-updates keep your node current with one click.
Earnings: $0.60/hr (80% of $0.75 client rate)
Operators receive 80% of client rate. Per-second billing.
Become a Node OperatorOr email us directly: signup@simulacrumlabs.com
All Simulacrum compute is US-only sovereign infrastructure by default. Every job runs in a hardened Docker container with capability dropping, resource isolation, ClamAV malware scanning, Proof of Compute verification, secure wipe, and append-only audit logging. Your data never leaves American soil.
Contact us for enterprise and defense pricingOr email us directly: contact@simulacrumlabs.com