Now Online: 124 H100 Nodes

Open Weights
50% Cheaper

Stop overpaying AWS, Azure, and Runpod. Rent high-performance GPUs from our decentralized network. Optimized for Llama 3, Mixtral, and Stable Diffusion.

SUPPORTED MODELS: LLAMA-3-70B /// MIXTRAL-8X7B /// STABLE-DIFFUSION-XL /// QWEN-1.5 /// FALCON-180B /// MISTAL-LARGE /// SUPPORTED MODELS: LLAMA-3-70B /// MIXTRAL-8X7B /// STABLE-DIFFUSION-XL /// QWEN-1.5

Market Price Comparison

Real-time cost analysis based on hourly on-demand rates.

DURATION 100 HOURS
ESTIMATED TOTAL COST
$199 SUILO PRICE
vs AWS/Azure: $450
AWS / Azure (P5 Instances) $4.50/hr
Runpod / Lambda $2.89/hr
SUILO DECENTRALIZED $1.99/hr
Best Value

* Comparison based on average spot pricing availability.

Available Instances

/// INFERENCE OPTIMIZED

RTX 4090 (24GB)

Ideal for SDXL, Llama-3-8b, and testing.

$0.29/hr
  • > 1,008 GB/s Bandwidth
  • > PCIe Gen 4
  • > Instant Boot
Deploy
/// TRAINING STANDARD

NVIDIA A100 (80GB)

Best value for fine-tuning 70b models.

$0.99/hr
  • > 2 TB/s Bandwidth
  • > SXM4 & NVLink
  • > 600 GB/s Interconnect
Deploy
/// FOUNDATION MODELS

NVIDIA H100 (80GB)

Extreme scale training clusters.

$1.99/hr
  • > 3.35 TB/s Bandwidth
  • > FP8 Transformer Engine
  • > 3.2Tbps Infiniband
Contact Sales

Start Training

Access is currently granted in waves to ensure network stability.