V 2.4.1 // Swarm Intelligence

EDGE
COMPUTE.

Decentralized network of 140,000+ GPUs. Rent dynamic edge capacity for AI inference and distributed model training.

Network Capacity
84,205 TFLOPs
Active Nodes
142,891
Swarm Training POPULAR

Cluster Setup

Distributed PyTorch/Ray training across heterogeneous A100/H100 swarms.

  • Fault Tolerance     Auto-heal
  • Scaling     Elastic
$0.45 / TFLOP-Hr
Edge Inference

API Routing

Geographically routed inference for LLMs on RTX 4090 networks.

$0.0002
Per 1K Tokens
# Connect to Edge Node
import edgegrid

client = edgegrid.Swarm(api_key="EDGE_KEY")

# Submit distributed job
job = client.train.submit(
    model="llama-3",
    strategy="pipeline_parallel",
    max_nodes=128
)
System updated.