Compute Comparison

Compare listed CoreWeave GPU hourly rates, models and regions below. Check the provider for availability, full instance cost and billing terms before choosing.

CoreWeave

Specialist Cloud

GB300 NVL72, HGX B200, HGX H200 — Gartner Visionary 2026

CoreWeave is the world's leading AI-native cloud, recognized as a Visionary in the 2026 Gartner® Magic Quadrant™ for Cloud AI Infrastructure and the only AI cloud to earn SemiAnalysis Platinum ClusterMAX™ twice. GB300 NVL72, GB200 NVL72, HGX B300, HGX B200, RTX PRO 6000, GH200, HGX H200, HGX H100, L40S, A100 — also available in Europe. Spot pricing available at 40–60% discount. NVIDIA Vera Rubin NVL72 available. Zero-egress data migration (CoreWeave 0EM). Trusted by OpenAI, Mistral AI, IBM, Jane Street, Aston Martin Aramco F1. Up to 47% TCO savings on 3-year capacity plans.

Visit CoreWeave

Cheapest On-Demand CoreWeave Pricing

$0.22/hr

Cheapest Spot CoreWeave Pricing

$0.07/hr

CoreWeave GPU Listings

9

CoreWeave Billing

On-demand, Spot, Reserved, Capacity plans (1yr/3yr)

CoreWeave Performance Benchmarks

Loading…

Compare CoreWeave With Another Cloud Provider

CoreWeave Info

Headquarters

Livingston, NJ

Founded

2017

Regions

US-East, US-West, EU-West

Min Commitment

None (on-demand)

Support

Standard → Enterprise

CoreWeave Strengths

  • ▸GB300 NVL72 and GB200 NVL72 — latest Vera Rubin and Blackwell hardware
  • ▸HGX B200 and HGX H200 — latest Blackwell and Hopper nodes
  • ▸Spot pricing at 40–60% off on-demand across all GPU types
  • ▸Zero-egress data migration (CoreWeave 0EM)
  • ▸Gartner Visionary 2026 — only AI cloud with Platinum ClusterMAX™
  • ▸Trusted by OpenAI, Mistral AI, IBM, Jane Street

CoreWeave Weaknesses

  • ▸Sales-led process for large clusters — not fully self-serve
  • ▸Fewer regions than hyperscalers (US and EU only)
  • ▸Minimum spend requirements for enterprise contracts

CoreWeave is Best For

Frontier model training at hyperscaleProduction LLM inference at low latencyAI labs and enterprises needing maximum cluster reliabilityTeams cutting hyperscaler costs — up to 47% TCO savings on 3-year plans

CoreWeave GPU Catalog

← swipe to scroll →
GPU ModelvRAMOn-DemandSpotAvailabilityRegionSource
RTX 309024 GB$0.22$0.07Medus-east-1live APIretrieved 8m ago
RTX 409024 GB$0.44$0.13Highus-east-1live APIretrieved 8m ago
A4048 GB$0.65$0.20Highus-east-1live APIretrieved 8m ago
A100 40GB40 GB$1.09$0.33Medus-east-1live APIretrieved 8m ago
L4048 GB$1.25$0.78Medus-east-1live APIretrieved 8m ago
L40S48 GB$2.25$0.99Medus-east-1live APIretrieved 8m ago
A100 80GB80 GB$2.70$1.21Highus-east-1live APIretrieved 8m ago
H100 80GB80 GB$6.16$2.46Highus-east-1live APIretrieved 8m ago
H200 141GB141 GB$6.30$2.62Medus-east-1live APIretrieved 8m ago

CoreWeave Reviews

4.3(6)
4.3
6 reviews
5
3
4
2
3
1
2
0
1
0
Sort:
Most helpful review
Best H100 availability we've found at scale
LLM pre-training (64× H100 SXM)Sep 2026

We run 64× H100 SXM clusters for LLM pre-training and CoreWeave has been rock-solid. Provisioning is fast — usually under 10 minutes for reserved capacity. The Kubernetes-native setup integrates cleanly with our existing Argo Workflows pipelines. Networking is genuinely fast; we see ~3,100 Gbps aggregate across a full pod. Support responds within the hour on Slack. Pricing is meaningfully cheaper than AWS P5 for equivalent hardware.

Anonymous reviewer
Great for inference, onboarding takes effort
vLLM inference servingSep 2026

Deployed a multi-replica vLLM inference stack on CoreWeave. Once running, latency and throughput are excellent — better than any other cloud we tested for sustained inference load. The onboarding process requires a sales call and credit check which slowed us down by a week. Not ideal for startups that want instant access. Once you're in, though, the platform is genuinely excellent.

Anonymous reviewer
Infiniband makes a measurable difference
TrainingAug 2026

Ran the same distributed training job on CoreWeave and a major cloud. CoreWeave finished 23% faster on identical GPU counts. The Infiniband interconnect is not marketing — it genuinely matters for multi-node runs.

Anonymous reviewer
Great infra, complex onboarding
DevelopmentAug 2026

CoreWeave's infrastructure is genuinely excellent but the onboarding process is more involved than other providers — Kubernetes knowledge is basically required.

Anonymous reviewer
Excellent for large-scale training runs
TrainingAug 2026

Used CoreWeave for a 512-GPU training run. Infiniband networking made a real difference vs. other providers. Pricing is higher than RunPod but the reliability justifies it.

Anonymous reviewer
1 / 2

CoreWeave — Frequently Asked Questions (FAQs)

Browse Other GPU Providers

CoreWeave GPU pricing overview

CoreWeave is a specialist GPU cloud provider headquartered in Livingston, NJ. Billing is On-demand, Spot, Reserved, Capacity plans (1yr/3yr) with a minimum commitment of None (on-demand). Available regions include US-East, US-West, EU-West. GPU instances can be provisioned on-demand, making CoreWeave suitable for both short-duration experiments and sustained production workloads.

CoreWeave vs other GPU providers

CoreWeave competes with providers including Lambda Labs, RunPod, Paperspace, Vast.ai, and the major hyperscalers (AWS, Google Cloud, Azure) for GPU compute workloads spanning LLM training, fine-tuning, and inference serving. Key differentiators include: GB300 NVL72 and GB200 NVL72 — latest Vera Rubin and Blackwell hardware; HGX B200 and HGX H200 — latest Blackwell and Hopper nodes; Spot pricing at 40–60% off on-demand across all GPU types. Use the side-by-side comparison tool above to see CoreWeave pricing against any other provider across shared GPU models. For a broader market view, the live GPU prices table shows all 9 CoreWeave listings alongside 91+ providers in a single sortable view.

Best use cases for CoreWeave

CoreWeave is best suited for: Frontier model training at hyperscale, Production LLM inference at low latency, AI labs and enterprises needing maximum cluster reliability, Teams cutting hyperscaler costs — up to 47% TCO savings on 3-year plans. Support tiers range from Standard → Enterprise, making it viable for both individual researchers and enterprise teams with SLA requirements. There are currently 9 active GPU listings on CoreWeave, covering H100 80GB, H200 141GB, A100 80GB, A100 40GB and more. H100 SXM5 instances with NVLink interconnect are commonly used for large-scale training workloads. A100 40GB and RTX 4090 instances are frequently used for fine-tuning and inference of smaller models.

CoreWeave billing model and cost structure

CoreWeave uses On-demand, Spot, Reserved, Capacity plans (1yr/3yr) pricing. Spot (interruptible) instances are available from $0.07/hr — interruptible instances are typically lower-cost than on-demand, suitable for fault-tolerant training jobs with checkpointing. Reserved instance pricing may be available for predictable long-running workloads — check the provider's pricing page for current committed-use rates. Always compare the effective hourly rate including egress, storage, and networking costs when evaluating total cost of ownership across providers.

Choosing the right GPU on CoreWeave

GPU selection depends on model size, precision, and whether your workload is compute-bound or memory-bandwidth-bound. For LLM training above 30B parameters, H100 80GB SXM5 instances with NVLink are the standard choice — high HBM3 bandwidth and FP16 throughput make them well-suited for large transformer workloads. For inference of 7B–13B models in FP16 or BF16, A100 40GB is a common choice on many providers. RTX 4090 instances are widely used for fine-tuning, prototyping, and quantized inference (INT4/INT8) of models up to 70B. Read the H100 vs A100 guide or the GPU benchmarks for ML guide for a full breakdown.

How CoreWeave pricing data is collected

Prices shown are sourced from CoreWeave's public pricing API or pricing page and refreshed every 15 minutes. On-demand rates reflect the current list price for a single GPU instance in the cheapest available region. Spot prices, where available, reflect interruptible instance rates at the time of the last snapshot. All prices are in USD per hour. Daily snapshots are kept in our history archive and visualised in the GPU price history charts — useful for identifying seasonal pricing patterns and evaluating whether current rates are above or below the 30-day average.

Evaluating managed LLM inference APIs as an alternative to self-hosted GPU compute? Compare live LLM token prices across OpenAI, Anthropic, Google, Groq, and many other providers. The cheapest GPU cloud guide covers the break-even analysis between self-hosted and managed inference at different request volumes.

Ready to get started?

Launch your first GPU on CoreWeave

On-demand from $0.22/hr — 9 GPU configurations available. On-demand, Spot, Reserved, Capacity plans (1yr/3yr) billing.

Go to CoreWeave

Looking for alternatives?

Compare CoreWeave with other GPU cloud providers side by side.

CoreWeave alternatives
CoreWeave
Visit CoreWeave