GPU and model API pricing
GPUs are billed by the minute; model APIs are billed per token, on one prepaid balance.
Per-minute billing
Destroy the instance any time; unused minutes go back to your balance.
About 2-minute deploy
Pick a template, click Deploy. Root SSH and a public HTTPS endpoint included.
| GPU Model | CloudGPU |
|---|---|
| RTX 5090 32G | $0.59/hr |
| RTX 4090 48GSame chip as the 4090, double the VRAM | $0.65/hr |
| RTX 4090 24G | $0.33/hr |
| RTX 3090 24G | $0.21/hr |
| V100 32G | $0.24/hr |
| A100 80G | $1.49/hr |
| H20 96GHopper with reduced compute, large HBM3 | $1.29/hr |
| Ascend 910B 64GNPU — CANN / MindSpore / PyTorch-NPU only | $0.95/hr |
| A800 80GA100 variant, NVLink capped at 400 GB/s | $1.09/hr |
| A800 40GA100 variant, NVLink capped at 400 GB/s | $0.69/hr |
| H800H100 variant, NVLink capped at 400 GB/s | $1.79/hr |
| RTX 2080 | $0.15/hr |
| A30 | $0.22/hr |
| RTX 4090D | $0.32/hr |
| A40 | $0.39/hr |
| RTX A6000 | $0.44/hr |
| A100 40G | $0.63/hr |
| RTX 6000D | $1.29/hr |
| RTX PRO 6000 96G | $1.49/hr |
CloudGPU is in public beta. There is no SLA: no uptime guarantee and no service credits. Unused GPU minutes go back to your balance when you destroy the instance, and the balance does not expire.
Performance
Price alone doesn't tell you which GPU is right. L20, H20, and Ascend 910B have distinctive VRAM and bandwidth profiles worth comparing.
| GPU | Arch | VRAM | FP16 TFLOPS | Mem BW |
|---|---|---|---|---|
RTX 5090 32G | Blackwell | 32 GB | 209 | 1792 GB/s |
RTX 4090 48G Same chip as the 4090, double the VRAM | Ada | 48 GB | 165 | 1008 GB/s |
RTX 4090 24G | Ada | 24 GB | 165 | 1008 GB/s |
RTX 3090 24G | Ampere | 24 GB | 71 | 936 GB/s |
V100 32G | Volta | 32 GB | 125 | 900 GB/s |
A100 80G | Ampere | 80 GB | 312 | 1935 GB/s |
H20 96G Hopper with reduced compute, large HBM3 | Hopper | 96 GB | 148 | 4000 GB/s |
Ascend 910B 64G NPU — CANN / MindSpore / PyTorch-NPU only | Ascend | 64 GB | 320 | 1200 GB/s |
A800 80G A100 variant, NVLink capped at 400 GB/s | Ampere | 80 GB | 312 | 2039 GB/s |
A800 40G A100 variant, NVLink capped at 400 GB/s | Ampere | 40 GB | 312 | 1555 GB/s |
H800 H100 variant, NVLink capped at 400 GB/s | Hopper | 80 GB | 989 | 3350 GB/s |
VRAM need depends on model, precision and workflow: How much VRAM do I need to run an LLM?
Cost Calculator
Pricing Plans
On-Demand
Pay as you go. No commitment.
Hourly list price
- check Per-minute billing
- check Cancel anytime
- check No minimum spend
- check All GPU models
Prepaid Credits
- check One top-up of $10 or more: $5 bonus, credited once you've spent your own $10. Once per account.
- check Credits never expire
- check Use on any GPU model
- check Works on GPUs and the model API
Enterprise
Steady volume, its own quote.
Talk to us
Priced on your volume
- check Reserved capacity
- check Volume rates
- check Multi-card clusters
- check A named contact
No hidden fees · No egress charges · Per-minute billing · Cancel anytime
Claude model rates
Our per-1M-token rates for the Claude models on CloudGPU, next to the official list prices. Official list prices checked 2026-10-24 on the official Claude pricing page.
| Model | CloudGPU input / output, per 1M tokens | Official list price, input / output, per 1M tokens | CloudGPU cached input, per 1M tokens | Official cache-hit list price, per 1M tokens |
|---|---|---|---|---|
Claude Opus 5.5 claude-opus-5-5 | $2.00 / $10.00 | $4.00 / $20.00 | $0.10 | $0.20 |
Claude Opus 5 claude-opus-5 | $2.50 / $12.50 | $5.00 / $25.00 | $0.25 | $0.50 |
Claude Opus 4.8 claude-opus-4-8 | $2.50 / $12.50 | $5.00 / $25.00 | $0.25 | $0.50 |
Claude Opus 4.7 claude-opus-4-7 | $2.50 / $12.50 | $5.00 / $25.00 | $0.25 | $0.50 |
Claude Opus 4.6 claude-opus-4-6 | $2.50 / $12.50 | $5.00 / $25.00 | $0.25 | $0.50 |
Claude Sonnet 5.5 claude-sonnet-5-5 | $1.00 / $5.00 | $2.00 / $10.00 | $0.05 | $0.10 |
Claude Sonnet 5 claude-sonnet-5 | $1.00 / $5.00 | $2.00 / $10.00 | $0.10 | $0.20 |
Claude Sonnet 4.6 claude-sonnet-4-6 | $1.50 / $7.50 | $3.00 / $15.00 | $0.15 | $0.30 |
Claude Fable 5.1 claude-fable-5-1 | $5.00 / $25.00 | $10.00 / $50.00 | $0.125 | $0.25 |
Claude Fable 5 claude-fable-5 | $5.00 / $25.00 | $10.00 / $50.00 | $0.50 | $1.00 |
Claude Opus 5.5 claude-opus-5-5
CloudGPU, per 1M tokens
input $2.00 / output $10.00 / cached input $0.10
Official list price, per 1M tokens
input $4.00 / output $20.00 / cache hits $0.20
Claude Opus 5 claude-opus-5
CloudGPU, per 1M tokens
input $2.50 / output $12.50 / cached input $0.25
Official list price, per 1M tokens
input $5.00 / output $25.00 / cache hits $0.50
Claude Opus 4.8 claude-opus-4-8
CloudGPU, per 1M tokens
input $2.50 / output $12.50 / cached input $0.25
Official list price, per 1M tokens
input $5.00 / output $25.00 / cache hits $0.50
Claude Opus 4.7 claude-opus-4-7
CloudGPU, per 1M tokens
input $2.50 / output $12.50 / cached input $0.25
Official list price, per 1M tokens
input $5.00 / output $25.00 / cache hits $0.50
Claude Opus 4.6 claude-opus-4-6
CloudGPU, per 1M tokens
input $2.50 / output $12.50 / cached input $0.25
Official list price, per 1M tokens
input $5.00 / output $25.00 / cache hits $0.50
Claude Sonnet 5.5 claude-sonnet-5-5
CloudGPU, per 1M tokens
input $1.00 / output $5.00 / cached input $0.05
Official list price, per 1M tokens
input $2.00 / output $10.00 / cache hits $0.10
Claude Sonnet 5 claude-sonnet-5
CloudGPU, per 1M tokens
input $1.00 / output $5.00 / cached input $0.10
Official list price, per 1M tokens
input $2.00 / output $10.00 / cache hits $0.20
Claude Sonnet 4.6 claude-sonnet-4-6
CloudGPU, per 1M tokens
input $1.50 / output $7.50 / cached input $0.15
Official list price, per 1M tokens
input $3.00 / output $15.00 / cache hits $0.30
Claude Fable 5.1 claude-fable-5-1
CloudGPU, per 1M tokens
input $5.00 / output $25.00 / cached input $0.125
Official list price, per 1M tokens
input $10.00 / output $50.00 / cache hits $0.25
Claude Fable 5 claude-fable-5
CloudGPU, per 1M tokens
input $5.00 / output $25.00 / cached input $0.50
Official list price, per 1M tokens
input $10.00 / output $50.00 / cache hits $1.00
Official list prices can change after the check date; the official pricing page is the source for them. Our rates come from the live price table on /api. CloudGPU is an independent service, not the official Claude API.
How You Pay
Add credit to your account, then spend it on GPU rentals or the AI Gateway. One balance covers both. GPU time is billed by the minute; API calls by the token.
Debit or credit card
Alipay, Google Pay, Apple Pay, Visa and Mastercard in one secure checkout, any amount from $5. Handled by Creem as merchant of record; the charge shows as CREEM.IO*CLOUDGPU.APP.
Choose Card on the Billing page and enter an amount. The balance is credited automatically once the payment is confirmed.
Bank Transfer
USD, EUR, GBP and 8 more currencies. From $50.
We give you our account details and an amount with a unique cent value, which is how your transfer is matched to your account. Usually 1–2 business days.
USDT (TRC-20)
Tether on the Tron network, credited 1:1.
Each account gets its own deposit address. Send from any wallet or exchange; the balance updates once the transfer is confirmed on-chain.
All prices are in US dollars. No setup fee, no egress charge, no minimum spend.
Frequently asked questions
How is GPU time billed?
Per minute, rounded up, with a 1-minute minimum. One hour is pre-authorised while an instance runs, and unused minutes go back to your balance when you destroy the instance. There is no subscription and no minimum term.
How are model API calls billed?
Per token for chat models, and per image, per character or per second for image, speech and transcription models. The current rate for every model is listed at https://cloudgpu.app/api. Calls are paid from the same prepaid balance as GPU time.
Is there a subscription or minimum spend?
No. There is no subscription and no minimum term. GPU time is paid per minute and model calls per token, from a prepaid balance.
How can I pay?
Top up one prepaid balance by card or with USDT on TRC-20. Businesses can also pay by bank transfer. Methods and minimums are listed at https://cloudgpu.app/payments.
Does my balance expire?
No. The prepaid balance does not expire and cannot be transferred between users.
What happens to my data when I destroy an instance?
Destroying an instance deletes everything stored on it, and an instance is destroyed if the balance runs out. There is no persistent storage across instances yet, so copy your results off the instance before you destroy it.