BOOKCOMPUTE.com
Console →
The operating system for AI compute

Describe your
workload.

Stop searching GPUs. Tell BookCompute what you're building. Our advisor scans 120+ providers, compares live pricing, and returns a reservation-ready recommendation.

AI
⌘ / Ctrl + Enter to advise
Try:
H100/AWS$2.33▼ 5.0%|H100/AZR$2.39▼ 4.2%|H100/GCP$2.45▼ 3.4%|H100/CWV$2.51▼ 5.0%|H100/LMB$2.57▼ 4.2%|B200/AWS$4.68▼ 0.8%|B200/AZR$4.74▲ 0.0%|B200/GCP$4.80▲ 0.8%|B200/CWV$4.86▼ 0.8%|B200/LMB$4.92▲ 0.0%|RTX/AWS$0.73▲ 0.4%|RTX/AZR$0.79▲ 1.2%|RTX/GCP$0.85▲ 2.0%|RTX/CWV$0.91▲ 0.4%|RTX/LMB$0.97▲ 1.2%|TPU/AWS$2.98▼ 2.3%|TPU/AZR$3.04▼ 1.5%|TPU/GCP$3.10▼ 0.7%|TPU/CWV$3.16▼ 2.3%|TPU/LMB$3.22▼ 1.5%|H200/AWS$2.86▲ 0.0%|H200/AZR$2.92▲ 0.8%|H200/GCP$2.98▲ 1.6%|H200/CWV$3.04▲ 0.0%|H200/LMB$3.10▲ 0.8%|A100/AWS$1.30▼ 2.9%|A100/AZR$1.36▼ 2.1%|A100/GCP$1.42▼ 1.3%|A100/CWV$1.48▼ 2.9%|A100/LMB$1.54▼ 2.1%|MI300X/AWS$2.08▼ 1.6%|MI300X/AZR$2.14▼ 0.8%|MI300X/GCP$2.20▲ 0.0%|MI300X/CWV$2.26▼ 1.6%|MI300X/LMB$2.32▼ 0.8%|H100/AWS$2.33▼ 5.0%|H100/AZR$2.39▼ 4.2%|H100/GCP$2.45▼ 3.4%|H100/CWV$2.51▼ 5.0%|H100/LMB$2.57▼ 4.2%|B200/AWS$4.68▼ 0.8%|B200/AZR$4.74▲ 0.0%|B200/GCP$4.80▲ 0.8%|B200/CWV$4.86▼ 0.8%|B200/LMB$4.92▲ 0.0%|RTX/AWS$0.73▲ 0.4%|RTX/AZR$0.79▲ 1.2%|RTX/GCP$0.85▲ 2.0%|RTX/CWV$0.91▲ 0.4%|RTX/LMB$0.97▲ 1.2%|TPU/AWS$2.98▼ 2.3%|TPU/AZR$3.04▼ 1.5%|TPU/GCP$3.10▼ 0.7%|TPU/CWV$3.16▼ 2.3%|TPU/LMB$3.22▼ 1.5%|H200/AWS$2.86▲ 0.0%|H200/AZR$2.92▲ 0.8%|H200/GCP$2.98▲ 1.6%|H200/CWV$3.04▲ 0.0%|H200/LMB$3.10▲ 0.8%|A100/AWS$1.30▼ 2.9%|A100/AZR$1.36▼ 2.1%|A100/GCP$1.42▼ 1.3%|A100/CWV$1.48▼ 2.9%|A100/LMB$1.54▼ 2.1%|MI300X/AWS$2.08▼ 1.6%|MI300X/AZR$2.14▼ 0.8%|MI300X/GCP$2.20▲ 0.0%|MI300X/CWV$2.26▼ 1.6%|MI300X/LMB$2.32▼ 0.8%|
Live Market

Real-time GPU depth, every provider.

Streaming · 14ms latency
NVIDIA · Hopper
H100 SXM5
▼ 4.2%
$2.45/gpu/hr
1,402 units82% avail
NVIDIA · Blackwell
B200 NVL72
—
$4.80/gpu/hr
Ships Q3 · reservations open12% avail
NVIDIA · Ada
RTX PRO 6000
▲ 1.2%
$0.85/gpu/hr
8,920 units98% avail
Google · Google TPU
TPU v5p
▼ 1.5%
$3.10/gpu/hr
512 units62% avail
NVIDIA · Hopper
H200 SXM
▲ 0.8%
$2.98/gpu/hr
640 units54% avail
NVIDIA · Ampere
A100 80GB
▼ 2.1%
$1.42/gpu/hr
4,820 units88% avail
AMD · AMD Instinct
MI300X
▼ 0.8%
$2.20/gpu/hr
320 units41% avail
Provider Marketplace

120+ providers. One reservation surface.

From hyperscalers to neoclouds and spot networks. Compare inventory, uptime and region coverage — reserve in a single contract.

View directory →
CodeProviderInventoryUptimeFrom
  • AWS
    Amazon Web Services
    Hyperscaler · 32 regions
    82,400
    99.99%
    From $2.24
  • AZR
    Microsoft Azure
    Hyperscaler · 28 regions
    61,200
    99.98%
    From $2.35
  • GCP
    Google Cloud
    Hyperscaler · 24 regions
    54,800
    99.99%
    From $2.29
  • CWV
    CoreWeave
    Tier 1 · 9 regions
    45,200
    99.99%
    From $2.12
  • LMB
    Lambda Cloud
    Tier 1 · 5 regions
    12,400
    99.99%
    From $1.85
  • RPD
    RunPod
    Neocloud · 12 regions
    8,100
    99.95%
    From $1.65
  • NBS
    Nebius
    Tier 1 · 4 regions
    22,000
    99.98%
    From $2.00
  • OVH
    OVHcloud
    Tier 1 · 14 regions
    7,800
    99.95%
    From $2.40
  • VST
    Vast.ai
    Spot · 40 regions
    18,500
    99.85%
    From $0.98
Why BookCompute

An institutional layer for a fragmented market.

01

One Search

Structured query across every provider, chip class and region. Booking-com density for infrastructure.

02

All Providers

Hyperscalers, neoclouds, sovereign clouds and spot networks — unified with a single API.

03

Live Prices

Real-time bid/ask, price history and forecasts. Buy at the right moment, on the right venue.

04

Enterprise Ready

MSA, invoicing, SOC2, sovereign filters. Reserve 500+ GPU clusters with a single counterparty.

05

Global Marketplace

Match supply and demand across 40 countries. Resell idle reserved capacity instantly.

Enterprise RFQ

Request 500 GPUs.
Receive private bids.

Submit a structured request. Tier-1 providers respond within 4 hours with pricing, topology, and reservation windows. Awarded in 24h.

Open RFQ desk →Enterprise portal
Recent requestsBidsBest
  • 512× H100 · EU-West · 90 days7$2.08/hr
  • 1024× H200 · US-East · 6 months5$2.71/hr
  • 128× B200 · Reserved 20263$4.42/hr
Spot Marketplace

Last-minute compute.

4 live
H100 · CoreWeave
US-East
2h 14m
$1.62
−34%
A100 · Lambda
US-West
5h 40m
$0.71
−28%
H200 · Nebius
EU-West
18m
$2.44
−22%
RTX PRO · Vast
Global
1d 3h
$0.42
−51%
Price Intelligence

Every SKU, every venue, every hour.

Historical prices, cross-provider spreads, and a 30-day forecast built on 3.4B in transacted compute. Set alerts on the levels that matter.

30D spread
−12.4%
Best venue
CoreWeave
Forecast H100
$2.12/hr
Advisory
Buy spot
90-Day price history · $/GPU/hr
H100 SXM5A100 80GBB200 (fwd)
Benchmark Center

Compare compute like an analyst.

Latency, training throughput, inference tokens/sec, and interconnect fabric — measured on identical workloads across providers.

ProviderChipsetLatency P99Training TF/sInference tok/sInterconnect$/hr
CoreWeave8× H100 SXM514ms
989
12,400
3.2 Tb/s NVLink$18.40
Lambda8× H100 SXM516ms
972
12,100
3.2 Tb/s NVLink$17.90
Nebius8× H20012ms
1041
14,200
3.2 Tb/s NVLink$22.60
AWS8× H100 P518ms
951
11,800
3.2 Tb/s EFA$20.80
Google CloudTPU v5p-811ms
918
15,300
OCS Fabric$19.40
RunPod8× A100 80GB28ms
468
6,400
600 GB/s NVLink$9.60
$0.0B
Compute compared
0
Providers indexed
0
Countries covered
0.00%
Platform uptime
Ready?

Optimize your AI infrastructure.

Search 120+ providers in one place. Reserve capacity, receive private bids, and settle in a single contract.