⚡Under $0.50/hr🧠VRAM Estimator⚖Compare GPUs🎁Free LLM APIs🎯Model Index
Market Telemetry2026-09-158 min read

Cloud GPU Pricing Index 2026: The Definitive Infrastructure Directory

Comprehensive market benchmark evaluating spot rates, reserved tiers, and true cost-per-FLOP across verified enterprise and neo-cloud providers.

Table of Contents

Market OverviewSpot vs Reserved MathHyperscaler PremiumCost Per FLOP Analysis

Live Cloud GPU Pricing Matrix

Real-time spot and reserved rates across all tracked providers.

ProviderGPU & VRAMInterconnectSpot RateOn-DemandMonthlyStatusAction
Community
PCIe 4.0 (64 GB/s)$0.34 / hr$0.85 / hr$208 / moInstant
Deploy →
Bare Metal
PCIe 4.0 (64 GB/s)$0.69 / hr$1.73 / hr$422 / moInstant
Deploy →
Community
PCIe 4.0 (64 GB/s)$0.69 / hr$1.73 / hr$422 / moInstant
Deploy →
Cloud
PCIe 4.0 (64 GB/s)$0.74 / hr$1.85 / hr$453 / moInstant
Deploy →
Dedicated
PCIe 4.0 (64 GB/s)$0.89 / hr$2.23 / hr$545 / moInstant
Deploy →
Cloud
PCIe 4.0 (64 GB/s)$1.09 / hr$2.73 / hr$667 / moInstant
Deploy →
Bare Metal
PCIe 4.0 (64 GB/s)$1.19 / hr$2.97 / hr$728 / moInstant
Deploy →
Dedicated
PCIe 4.0 (64 GB/s)$1.49 / hr$3.73 / hr$912 / moInstant
Deploy →
Dedicated
N/A$1.59 / hr$3.98 / hr$973 / moInstant
Deploy →
Community
NVLink 4.0 (900 GB/s)$1.89 / hr$4.72 / hr$1,157 / moInstant
Deploy →
Bare Metal
NVLink 4.0 (900 GB/s)$2.29 / hr$5.73 / hr$1,401 / moInstant
Deploy →
Community
NVLink 4.0 (900 GB/s)$2.79 / hr$6.98 / hr$1,707 / moInstant
Deploy →
Dedicated
NVLink 4.0 (900 GB/s)$2.99 / hr$7.48 / hr$1,830 / moInstant
Deploy →
Bare Metal
NVLink 4.0 (900 GB/s)$3.19 / hr$7.98 / hr$1,952 / moInstant
Deploy →
Cloud
NVLink 4.0 (900 GB/s)$3.49 / hr$8.73 / hr$2,136 / moInstant
Deploy →
Community
NVLink 5.0 (1.8 TB/s)$3.99 / hr$9.98 / hr$2,442 / moInstant
Deploy →
Dedicated
NVLink 4.0 (900 GB/s)$3.99 / hr$9.98 / hr$2,442 / moInstant
Deploy →
Cloud
NVLink 4.0 (900 GB/s)$4.31 / hr$10.77 / hr$2,638 / moInstant
Deploy →
Bare Metal
NVLink 5.0 (1.8 TB/s)$4.49 / hr$11.23 / hr$2,748 / moInstant
Deploy →
Dedicated
NVLink 5.0 (1.8 TB/s)$5.49 / hr$13.73 / hr$3,360 / moInstant
Deploy →
Cloud
NVLink 5.0 (1.8 TB/s)$5.99 / hr$14.98 / hr$3,666 / moInstant
Deploy →
Data Freshness: Public Cloud APIs & Market Scraping | Refreshed Daily (UTC)Benchmark Baseline: Ubuntu 24.04, CUDA 12.4, vLLM v0.6.x, PagedAttention v2, FlashAttention-3

Market Overview

The cloud GPU market in 2026 has bifurcated into two distinct tiers: hyperscalers (AWS, GCP, Azure) charging $6-8/hr for H100 access with enterprise SLAs, and specialist neo-clouds (Spheron, RunPod, Lambda Labs, Vast.ai) offering the same hardware at $1.89-$3.49/hr with varying infrastructure guarantees. The 130-260% hyperscaler premium buys VPC integration, IAM authentication, and compliance certifications — not better GPU performance. For teams without strict regulatory requirements, neo-clouds deliver identical FP8 TFLOPS at 55-73% lower cost.

Spot vs Reserved Math

Every cloud provider offers two pricing tiers: spot (variable, preemption risk) and reserved (fixed, committed). The math is simple: Monthly Cost (Spot) = hourly_spot × 720 hours Monthly Cost (Reserved) = hourly_spot × 0.85 × 720 hours The 0.85 multiplier reflects the standard ~15% discount for 1-month commitments. Break-even occurs at 320 hours/month for H100 — if you run 24/7 (720 hours), reserved saves $800+/month per GPU. For intermittent workloads (<200 hours/month), spot pricing is preferred.

Hyperscaler Premium Analysis

AWS P5 H100: $6.88/hr on-demand, ~$4.13/hr spot. GCP A3 H100: ~$3.82/hr on-demand, ~$2.29/hr spot. Specialist clouds: $1.89-$3.49/hr on-demand, $2.29/hr spot. The premium is not about GPU silicon — it's about ecosystem lock-in. AWS charges $0.09/GB egress, S3 storage fees, CloudWatch logging. Specialist clouds typically offer free egress and included monitoring. For checkpoint-heavy training, AWS egress can add 20-40% to total compute costs.

Cost Per FLOP Analysis

The true metric is cost per PetaFLOP-hour, not cost per GPU-hour. H100 at $1.89/hr delivers 1,979 FP8 TFLOPS = $0.96/PFLOP-hr. A100 at $1.59/hr delivers 624 FP8 TFLOPS = $2.55/PFLOP-hr. B200 at $5.99/hr delivers 2,250 FP8 TFLOPS = $2.66/PFLOP-hr. The H100 remains the cost-efficiency king for FP8 workloads. The B200's advantage emerges at FP4 (4,500 TFLOPS) where cost/PFLOP drops to $1.33 — but FP4 support is limited to Blackwell-native frameworks.
OR

OpenGPU Radar Systems Engineering Team

Independent compute telemetry and infrastructure analysis. Not affiliated with NVIDIA, cloud providers, or hardware vendors.

Related Guides