CoreWeave Pricing (2026)
Pure usage-based: you're billed per instance-hour on the hardware you spin up, with no seats, platform fees, or subscription tiers in the way. The value metric is GPU compute time, and the 'Custom/mo' placeholder is misleading since the page only lists hourly rates (e.g., HGX H100 at $49.24/hr per 8-GPU node, GB200 NVL72 at $42/hr).
- NVIDIA GB300 NVL72
- Contact Sales
- NVIDIA GB200 NVL72
- Contact Sales
- NVIDIA HGX B300
- Contact Sales
- NVIDIA HGX B200
- Contact Sales
- NVIDIA RTX PRO 6000 Blackwell Server Edition (High Memory)
- Contact Sales
- NVIDIA RTX PRO 6000 Blackwell Server Edition (Standard Memory)
- Contact Sales
- NVIDIA GH200
- Contact Sales
- NVIDIA HGX H100
- Contact Sales
- NVIDIA L40
- Contact Sales
- NVIDIA HGX H200
- Contact Sales
- NVIDIA L40S
- Contact Sales
- NVIDIA A100
- Contact Sales
Price watch
Get an email when CoreWeave changes its price.
No change recorded yet. You'll hear about the first one before most people notice.
One email per change, with the old and new price. No newsletter. Unsubscribe in one click.
NVIDIA GB300 NVL72
- GPU Count: 4
- VRAM (GB): 279
- vCPUs: 144
- System RAM (GB): 960
- Local Storage (TB): 61.44
NVIDIA GB200 NVL72
- GPU Count: 4
- VRAM (GB): 186
- vCPUs: 144
- System RAM (GB): 960
- Local Storage (TB): 30.72
NVIDIA HGX B300
- GPU Count: 8
- VRAM (GB): 270
- vCPUs: 192
- System RAM (GB): 4,096
- Local Storage (TB): 61.44
- Spot Price (Per Hour): $35.84
NVIDIA HGX B200
- GPU Count: 8
- VRAM (GB): 180
- vCPUs: 128
- System RAM (GB): 2,048
- Local Storage (TB): 61.44
- Spot Price (Per Hour): $34.11
NVIDIA RTX PRO 6000 Blackwell Server Edition (High Memory)
- GPU Count: 8
- VRAM (GB): 96
- vCPUs: 128
- System RAM (GB): 1,024
- Local Storage (TB): 7.68
- Spot Price (Per Hour): $11.09
NVIDIA RTX PRO 6000 Blackwell Server Edition (Standard Memory)
- GPU Count: 8
- VRAM (GB): 96
- vCPUs: 128
- System RAM (GB): 512
- Local Storage (TB): 7.68
- Spot Price (Per Hour): $9.56
NVIDIA GH200
- GPU Count: 1
- VRAM (GB): 96
- vCPUs: 72
- System RAM (GB): 480
- Local Storage (TB): 7.68
NVIDIA HGX H100
- GPU Count: 8
- VRAM (GB): 80
- vCPUs: 128
- System RAM (GB): 2,048
- Local Storage (TB): 61.44
- Spot Price (Per Hour): $19.71
NVIDIA L40
- GPU Count: 8
- VRAM (GB): 48
- vCPUs: 128
- System RAM (GB): 1,024
- Local Storage (TB): 7.68
- Spot Price (Per Hour): $6.27
NVIDIA HGX H200
- GPU Count: 8
- VRAM (GB): 141
- vCPUs: 128
- System RAM (GB): 2,048
- Local Storage (TB): 61.44
- Spot Price (Per Hour): $20.93
NVIDIA L40S
- GPU Count: 8
- VRAM (GB): 48
- vCPUs: 128
- System RAM (GB): 1,024
- Local Storage (TB): 7.68
- Spot Price (Per Hour): $7.88
NVIDIA A100
- GPU Count: 8
- VRAM (GB): 80
- vCPUs: 128
- System RAM (GB): 2,048
- Local Storage (TB): 7.68
- Spot Price (Per Hour): $9.65
Does CoreWeave have a free plan?
No. CoreWeave lists no free plan or trial on its pricing page. Pricing is custom; you'll need to talk to sales.
Want free? DigitalOcean has free plans: see all CoreWeave alternatives →
AI Pricing Analysis
Pricing Model
Pure usage-based: you're billed per instance-hour on the hardware you spin up, with no seats, platform fees, or subscription tiers in the way. The value metric is GPU compute time, and the 'Custom/mo' placeholder is misleading since the page only lists hourly rates (e.g., HGX H100 at $49.24/hr per 8-GPU node, GB200 NVL72 at $42/hr).
Tier Strategy
The 'tiers' are really GPU generations, not feature plans: L40/L40S and RTX PRO 6000 for inference and lighter workloads, A100/H100/H200 for mainstream training, and GB200/GB300/B200/B300 for frontier-scale runs. The upgrade trigger is model size and memory bandwidth, so teams graduate when their workload stops fitting on older silicon or when training time starts costing more than a pricier, faster node.
Competitive Positioning
CoreWeave plays the specialist premium-performance angle against hyperscalers and GPU neoclouds, leaning on early Blackwell access, NVL72 rack-scale systems, and bare-metal-style performance rather than winning on rock-bottom rates. They're courting AI labs and serious ML teams who care more about getting scarce cutting-edge capacity than shaving a few dollars off the hourly rate.
Growth Lever
There's no feature paywall; expansion is all about consuming more GPU-hours and moving up to newer, pricier hardware, with the 'Custom' framing signaling that big reserved-capacity commitments get negotiated off-list. Spot pricing, region choice, and the inference-platform-only single-GPU rates nudge customers toward deeper platform lock-in as workloads scale.
Want the strategist's read on the live page? Read the AI teardown of CoreWeave's pricing page →
CoreWeave Pricing FAQ
- How much does CoreWeave cost?
- Pricing is custom — you'll need to talk to their sales team for a quote.
- Is there a free plan?
- No free plan. You'll need to commit to a paid plan to get started.
- Can I try it before paying?
- Not at the moment. There's no free trial or free plan listed on their pricing page.
- How does the pricing work?
- Pure usage-based: you're billed per instance-hour on the hardware you spin up, with no seats, platform fees, or subscription tiers in the way. The value metric is GPU compute time, and the 'Custom/mo' placeholder is misleading since the page only lists hourly rates (e.g., HGX H100 at $49.24/hr per 8-GPU node, GB200 NVL72 at $42/hr).
- Which plan makes sense for me?
- The 'tiers' are really GPU generations, not feature plans: L40/L40S and RTX PRO 6000 for inference and lighter workloads, A100/H100/H200 for mainstream training, and GB200/GB300/B200/B300 for frontier-scale runs. The upgrade trigger is model size and memory bandwidth, so teams graduate when their workload stops fitting on older silicon or when training time starts costing more than a pricier, faster node.
More Cloud pricing
All Cloud pricing →-
AWS Pricing
Amazon's cloud computing platform.
See pricing -
DigitalOcean Pricing
Cloud infrastructure for developers.
From $0/mo -
Google Cloud Pricing
Google's cloud infrastructure and AI.
See pricing -
Hetzner Pricing
European cloud at honest prices.
From $5/mo -
Linode Pricing
Cloud infrastructure by Akamai.
From $5/mo -
Microsoft Azure Pricing
Microsoft's cloud platform.
See pricing -
OVHcloud Pricing
European cloud infrastructure.
See pricing -
Scaleway Pricing
European cloud provider.
See pricing
Weighing these against each other? Track CoreWeave and its alternatives → Pricing, features and reviews, in one weekly brief. Free for 14 days.
Comparing CoreWeave to something specific? Try CoreWeave vs AWS, CoreWeave vs DigitalOcean, or CoreWeave vs Google Cloud. Or see the best CoreWeave alternatives, ranked.
Set it up once. Get a plan every week.
Add the competitors you care about. Outmano keeps watching them, so you can stop checking their sites.
14-day free trial · $0 today