ComputeUnion — Real-time AI pricing

AI API pricing — save up to 90%.

Real-time pricing for thousands of AI models across 70+ cloud providers. GPU rental included. Updated daily.

Compare AI API pricesCompare GPU rentals

5813+

AI Models

52+

API Platforms

62

GPU Models

Save 99%

vs Official Price

Data Updated

AI API Price Comparison

Real-time comparison across providers — sorted cheapest first

Full page →
AI API price intelligenceFour categories, canonical models first, raw quotes second.

Language models

Flagship, recommended and mainstream vendors first.

All LLM →

Video generation

Readable model families first; raw provider names remain in full quotes.

All video →

Image generation

Readable image model families first.

View all →

Audio and speech

TTS, ASR and speech families first.

View all →

GPU Rental

On-demand every 24h · Spot every 6h

▶
H100 SXM580GB · 19 platforms
$1.60
▶
H100 PCIe80GB · 2 platforms
$2.49
▶
A100 80GB80GB · 14 platforms
$0.67
▶
L40S48GB · 10 platforms
$0.03
▶
RTX 409024GB · 4 platforms
$0.36
▶
H200 SXM5141GB · 10 platforms
$0.47
▶
T4 16GB16GB · 2 platforms
$0.59
▶
V100 32GB32GB · 6 platforms
$0.02
▶
RTX 6000 Ada48GB · 3 platforms
$0.30
▶
RTX 309024GB · 2 platforms
$0.20
▶
RTX 408016GB · 2 platforms
$0.10
▶
A40 48GB48GB · 6 platforms
$0.15
View all GPUs →

Latest Activities

View All →
Discountvast.ai

H100 Flash Sale — Select Instances Up to 30% Off

Limited H100 SXM5 80GB instances available at flash-sale pricing — up to 30% off market rate, from ~$1.74/hr. Available on a first-come first-served basis. New lots listed daily at random times.

Claim →

Frequently Asked Questions

What prices does ComputeUnion compare?

ComputeUnion currently tracks 5,813+ AI model prices across official APIs and relay providers, plus hourly GPU rental offers.

What is the difference between official API and relay pricing?

Official APIs come directly from model developers. Relays may offer lower prices or different payment options, but latency, SLA, data handling, and support can differ.

How often are prices updated?

Platforms with machine-readable pricing endpoints are auto-scraped. Other prices are maintained from official pricing pages, with update or maintenance status shown on the site.

When should I rent a GPU instead of buying an AI API?

APIs often fit fixed-model and lower-volume usage. GPU rental is worth comparing for self-hosting, sustained workloads, fine-tuning, or full control over model deployment.