System Status

All systems
operational.

Telemetry Pending
Live metrics coming soon
Active Nodes
—
across 5 regions
Running Jobs
—
at this moment
GPU Utilization
—
network average
Uptime
—
last 90 days
Active Endpoints
—
inference serving
Avg Job Boot
—
cold start time
Regional Network Map
US-EAST— nodes
——
US-WEST— nodes
——
EU-WEST— nodes
——
AP-SOUTH— nodes
——
AP-EAST— nodes
——
Coming Soon

Live telemetry not yet available

Real-time cluster metrics will be enabled in a future release.

Roadmap

Where we've been.
Where we're going.

From POC to full ecosystem — our development roadmap, phase by phase.

Phase 1

POC — Foundation

Summer 2025completed

Proof of concept built at ETHKyiv 2025. Core kernel, public GPU pool, basic training job submission. Won DeAI and State Channels tracks.

✓Public GPU pool (RTX 3090, RTX 4090)
✓Basic training job submission via platform
✓Per-account Garage S3 bucket with presigned-URL data access
✓Job status and logs
✓Pay-as-you-go billing
Phase 2

Beta Training

Fall — Winter 2025completed

Platform build-out and beta testing. First real training jobs ran on the cluster — DeepLab and image model fine-tuning. Zero lost checkpoints.

✓Multi-node distributed training
✓Automatic checkpoint recovery
✓DeepLab and image model fine-tuning
✓PyTorch, TensorFlow, HuggingFace support
✓REST API and Python SDK
✓Zero lost checkpoints across all beta runs
Phase 3IN PROGRESS

Managed Clusters + Inference — Beta

2026 — nowcurrent

Active beta testing of managed clusters and inference endpoints. Dedicated nodes, isolated kernel, and first live inference runs with early users.

→Bring your own model or select from the registry
→Dedicated, non-shared nodes (beta)
→Isolated job kernel
→Persistent state between runs
→Inference endpoints (beta)
→Custom GPU configurations
Phase 4

Inference Endpoints

Q3 2026upcoming

Production inference endpoints with autoscaling, OpenAI-compatible API, and Triton backends.

→OpenAI-compatible inference endpoints
→Triton inference server backend
→Autoscale to zero — pay per request
→Sub-10s cold starts on public pool
→Model registry with versioning
→Private Cluster tier (GA)
Phase 5

Full Ecosystem

H2 2026future

Complete end-to-end ML platform — dataset management, automated pipelines, model marketplace, full observability.

○Dataset management and versioning
○Automated retraining pipelines
○Full observability dashboard
○Cost optimization recommendations
○Federated learning support
○SOC 2 Type II certification

Join the ecosystem.
Run your first job today.

EXPLORE PLATFORM