// INFRASTRUCTURE INTELLIGENCE LAYER

The compute powering
every modern AI system.

Live benchmarks, uptime telemetry, and editorial rankings for 27+ GPU clouds, inference APIs, and hosting platforms , from Parasail to Oracle to Groq.

Providers tracked
27
live
Operational
27/27
last 60s
Median latency
124ms
cross-provider
Categories
10
benchmark suites

// TOP-OF-MARKET

Featured Providers

Editorial picks
hyperscaler

Amazon Web Services

The default cloud for AI workloads

Trust
95
Latency
55ms
Uptime
99.99%
H100/hr
$4.91
hyperscaler

Microsoft Azure

OpenAI's exclusive cloud

Trust
94
Latency
52ms
Uptime
99.98%
H100/hr
$4.50
inference-api

OpenAI API

The reference frontier API

Trust
94
Latency
420ms
Uptime
99.97%
H100/hr
,
hyperscaler

Google Cloud

TPUs and GPUs with Vertex AI

Trust
94
Latency
48ms
Uptime
99.99%
H100/hr
$3.67
gpu-cloud

NVIDIA DGX Cloud

First-party hyperscale GPU cloud

Trust
93
Latency
40ms
Uptime
99.99%
H100/hr
$4.95
inference-api

Anthropic API

Direct API to the Claude family

Trust
93
Latency
540ms
Uptime
99.96%
H100/hr
,

Fastest Inference

Tokens / second
01Fireworks AI2,400 tok/s02Together AI2,100 tok/s03Cerebras2,100 tok/s04Modal1,900 tok/s05Parasail1,850 tok/s

Cheapest H100

$ / GPU hour
01Vast.ai$1.6502Together AI$1.9903RunPod$1.9904Nebius$2.0005TensorDock$2.25

Best Uptime

Trailing 90d
01Amazon Web Services99.99%02Google Cloud99.99%03NVIDIA DGX Cloud99.99%04Oracle Cloud Infrastructure99.99%05Microsoft Azure99.98%

// LIVE RANKINGS

Infrastructure Trust Score

Composite of reliability, pricing fairness, speed, scalability, community reputation and editorial analysis.

#
Provider
Category
Trust
Speed
Latency
Uptime
H100 $/hr
Status
01
Amazon Web Services
The default cloud for AI workloads
hyperscaler
95
82
55ms
99.99%
$4.91
02
Microsoft Azure
OpenAI's exclusive cloud
hyperscaler
94
84
52ms
99.98%
$4.50
03
OpenAI API
The reference frontier API
inference-api
94
86
420ms
99.97%
,
04
Google Cloud
TPUs and GPUs with Vertex AI
hyperscaler
94
86
48ms
99.99%
$3.67
05
NVIDIA DGX Cloud
First-party hyperscale GPU cloud
gpu-cloud
93
89
40ms
99.99%
$4.95
06
Anthropic API
Direct API to the Claude family
inference-api
93
82
540ms
99.96%
,
07
Oracle Cloud Infrastructure
Enterprise AI compute at hyperscaler depth
hyperscaler
92
84
38ms
99.99%
$3.00
08
CoreWeave
The AI hyperscaler
gpu-cloud
91
90
42ms
99.95%
$2.39
09
Fireworks AI
Production inference for open models
inference-api
90
94
72ms
99.94%
,
10
Together AI
Open-source models, blazing fast
inference-api
89
93
98ms
99.92%
$1.99
11
Parasail
Serverless AI infrastructure for next-gen models
inference-api
88
92
142ms
99.93%
$2.49
12
Groq
LPUs for instant inference
inference-api
88
99
18ms
99.91%
,
13
Cerebras
Wafer-scale inference
inference-api
87
98
26ms
99.88%
,
14
Lambda Labs
Deep learning GPU cloud
gpu-cloud
86
88
46ms
99.90%
$2.49
15
OpenRouter
One API for 300+ models
inference-api
86
88
210ms
99.95%
,
16
Modal
Serverless Python for AI
gpu-cloud
86
89
84ms
99.94%
$3.95
17
Baseten
Production inference platform
inference-api
85
87
110ms
99.93%
$6.49
18
Hugging Face Inference
One API for every open model
inference-api
85
79
180ms
99.85%
$4.50
19
Crusoe Cloud
Climate-aligned GPU cloud
gpu-cloud
84
87
52ms
99.92%
$2.45
20
Fal.ai
Lightning-fast generative media
inference-api
84
95
42ms
99.91%
,
21
Nebius
European GPU hyperscaler
gpu-cloud
83
85
58ms
99.93%
$2.00
22
Replicate
Run any open-source model with an API
inference-api
83
76
210ms
99.82%
,
23
SambaNova
Datascale RDU inference
inference-api
83
93
42ms
99.85%
,
24
xAI API
Grok at frontier scale
inference-api
82
86
380ms
99.78%
,
25
RunPod
Affordable GPU containers on demand
gpu-cloud
79
84
68ms
99.70%
$1.99
26
TensorDock
Wholesale GPU cloud
gpu-cloud
72
78
142ms
99.00%
$2.25
27
Vast.ai
GPU marketplace at spot prices
gpu-cloud
71
76
130ms
98.50%
$1.65

// EDITORIAL RANKINGS

Best-Of Categories

#01 · trust_score

Best AI Inference Provider

Top-rated inference APIs ranked by composite Trust Score.

#02 · cheapest_h100_hourly

Cheapest GPU Cloud

Lowest hourly H100 pricing across major providers.

#03 · tokens_per_sec

Fastest Inference API

Highest tokens-per-second on Llama-3 70B class workloads.

#04 · reliability_score

Best Enterprise Infrastructure

SOC2/HIPAA/FedRAMP-grade providers with strongest SLAs.

#05 · community_score

Best Open-Source Infrastructure

Providers hosting the widest open-model catalog with best DX.

#06 · avg_latency_ms

Lowest Latency Provider

Median API response time at the edge.

#07 · uptime_pct

Best Uptime

Highest measured uptime over the trailing 90 days.

#08 · community_score

Best Developer Experience

Composite of docs quality, SDKs, OpenAI-compat, and community love.

#09 · editorial_score

Best AI Hosting Platform

Top-rated platforms for deploying custom and open models.

#10 · scalability_score

Best Model Deployment Platform

Platforms with the smoothest dedicated / fine-tuning workflow.