Price checkPer instance
On-Demand (8x GPU)CustomOn-Demand (1x GPU)CustomSpotCustom
See all 4 plans
Quick Answer
Last verified:
Medium confidence

CoreWeave costs $10 to $68.80 per instance/hour as of May 2026, with 4 plans available. The median contract is $726/year based on 34 verified purchases.

Use the interactive pricing calculator to estimate your exact cost based on team size and requirements.

  • Free tier: No free tier available

CoreWeave offers 4 pricing tiers: On-Demand (8x GPU), On-Demand (1x GPU), Spot, Reserved. The On-Demand (1x GPU) plan is inference and smaller workloads.

Compared to other ai/gpu cloud compute software, CoreWeave is positioned at the budget-friendly price point.

  • Median contract: $726/yr from 34 purchases
  • 5 documented hidden costs beyond list price
Last verified:

How much does CoreWeave cost?

CoreWeave pricing starts at $10/instance/hour across 4 plans, with enterprise pricing available on request. Plans include On-Demand (8x GPU) (custom pricing), On-Demand (1x GPU) (custom pricing), Spot (custom pricing), Reserved (custom pricing).

CoreWeave Pricing Overview

CoreWeave has 4 pricing plans ranging from $10 to $68.80/instance/hour. The On-Demand (8x GPU) plan requires contacting sales for a custom quote and is designed for large-scale ai training. The On-Demand (1x GPU) plan requires contacting sales for a custom quote and is designed for inference and smaller workloads. The Spot plan requires contacting sales for a custom quote and is designed for fault-tolerant batch workloads. The Reserved plan requires contacting sales for a custom quote and is designed for predictable long-term workloads.

The median CoreWeave customer pays $726/year based on 34 verified purchases.

There are at least 5 documented hidden costs beyond CoreWeave's list price, including implementation, training, and add-on fees.

This pricing was last verified in April 24, 2026.

See CoreWeave Plans

CoreWeave is a specialized GPU cloud provider offering four compute tiers — On-Demand (1x GPU), On-Demand (8x GPU), Spot, and Reserved — all at custom-quoted pricing negotiated through account managers. CoreWeave does not publish list prices; rates are determined by GPU model, configuration, and contract length. As a neocloud focused exclusively on AI infrastructure, CoreWeave is priced approximately 40-50% below AWS for comparable GPU instances, with Reserved contracts offering the lowest per-hour rates for teams with sustained compute needs.

How CoreWeave Pricing Compares

Compare CoreWeave pricing against top alternatives in AI/GPU Cloud Compute.

Lambda $0.69-$6.99/GPU/hour Compare → Hyperbolic $0.3-$3.2/GPU/hour Compare → RunPod $0.34-$3.49/GPU/hour Compare →

All CoreWeave Plans & Pricing

Plan Monthly Annual Best For
View all features by plan (compare side-by-side)

On-Demand (8x GPU)

  • HGX B200 (8 GPUs) at $68.80/hr
  • HGX H200 (8 GPUs) at $50.44/hr
  • HGX H100 (8 GPUs) at $49.24/hr
  • A100 (8 GPUs) at $21.60/hr
  • L40S (8 GPUs) at $18.00/hr
  • L40 (8 GPUs) at $10.00/hr
  • No ingress/egress fees

On-Demand (1x GPU)

  • GH200 (1 GPU) at $6.50/hr
  • RTX PRO 6000 Blackwell at $20.00/hr
  • GB200 NVL72 (4 GPUs) at $42.00/hr

Spot

  • RTX PRO 6000 Blackwell at $9.24/hr (spot)
  • Preemptible instances
  • Up to 54% savings

Reserved

  • Up to 60% discount over on-demand
  • Committed usage terms
  • Contact sales for pricing
See CoreWeave Plans
Compare CoreWeave with alternativesAdjust seats, lock a tier, add up to 2 more products side-by-side. Shareable URL.

Compare CoreWeave vs Alternatives

Before committing to CoreWeave, compare pricing with these 3 alternatives in the same category.

VSLambda
From $0.69/GPU/hour

Development, fine-tuning, and small inference jobs

Full comparison
VSHyperbolic
From $0.3/GPU/hour

Full comparison
VSRunPod
From $0.34/GPU/hour

Batch workloads, training runs, and cost-sensitive inference that can tolerate interruptions

Full comparison

All CoreWeave alternatives & migration guides

What Companies Actually Pay for CoreWeave

The median CoreWeave buyer pays $726/year based on 34 verified purchase transactions.

What companies actually pay $726/yr Median across 34 community cost mentions
Review scores
Top pricing complaints
No self-serve onboarding — requires meeting request and account manager approvalLarge GPU models not always available on-demandEnterprise focus means smaller customers receive less attention and supportVendor lock-in risk from proprietary orchestration tools (CKS, SUNK)
Source: Community cost mentions (Reddit, Hacker News) — aggregated from 34 distinct user reports. Indicative only — not contract-grade data.

How CoreWeave Pricing Compares

Software Starting Price Top Price
CoreWeave $10/instance/hour $68.8/instance/hour
Hyperbolic $0.3/GPU/hour $3.2/GPU/hour
Lambda $0.69/GPU/hour $6.99/GPU/hour
Modal Free $250/GPU/hour
Paperspace Free $39/GPU/hour
RunPod $0.34/GPU/hour $3.49/GPU/hour

Detailed pricing comparisons:

Browse all AI/GPU Cloud Compute pricing →

5 CoreWeave Hidden Costs Beyond the List Price

Beyond the listed price, CoreWeave has at least 5 documented hidden costs that can significantly increase total cost of ownership.

Watch for 5 hidden costs
  • Enterprise-Only Onboarding Friction 5-15% of license costs
    medium 1 source
    Reddit "CoreWeave's onboarding experience can feel a bit restrictive. New users must go through a 'Request a Meeting' page and fill out a contact form to outline their needs."
  • Reserved Contract Lock-In Risk 15-40% of license costs
    high 2 sources
    Reddit "CoreWeave... rent out compute on long-term contracts. But lately they've been leaning into the on-demand game too now that GPU supply has caught up and pricing's headed toward a race to zero on $ / GPU / hour."
    Reddit "Whether this relationship lasts, and whether Coreweave can keep it's pricing / cost advantage, is a another risk I see other than AI bubble bursting."
  • On-Demand GPU Availability Constraints 5-20% of license costs
    medium 2 sources
    Reddit "Coreweave and Llambda Labs (and Vultr looks like an option but I haven't tried them enough yet to give the thumbs up) - have good pricing but none of the "jank" of Vast rentals."
    Reddit "Coreweave and lambda labs are pretty good but the big gpus are not always available these days."
  • Vendor Lock-In via Proprietary Software Stack 10-25% of license costs
    medium 2 sources
    Reddit "the vendor lock-in is HARD on a lot of these things and if their pricing model changes I'm completely fucked, so I still end up going back to kubernetes and knative for running stuff more provider-agnostically."
    Reddit "Through partnership, acquisition and some in-house development, CoreWeave is expanding what it offers to improve its developer experience."
  • Reduced Attention for Smaller Customers 5-10% of license costs
    low 1 source
    Reddit "Not sure if they are accepting smaller time clients though these days - I can check with my account manager if you want…they are doing a lot of crazy things these days - mostly focused on building a ton of new data centers and AI data stuff so us ..."
Tip

Ask your CoreWeave sales rep about these costs upfront. Getting them in writing before signing can save you from surprise charges later.

Full hidden costs breakdown →

Intelligence sourced from 1 independent sources
Reddit User discussions
Key claims include inline source attribution. Data verified against multiple independent sources. 16 source citations total.

CoreWeave Contract Terms

CoreWeave contracts do not auto-renew. Changes require advance notice. These terms are sourced from verified buyer experiences.

Contract Terms
Auto-Renewal No
Minimum Commitment Long-term reserved contracts standard for enterprise customers; on-demand and spot available without commitment
Mid-Term Downgrade Not allowed
Payment Terms Long-term contracts standard for Reserved tier; on-demand hourly billing available
Price Escalation GPU compute pricing facing downward pressure industry-wide as supply normalizes; older GPU models (H100) seeing pricing pressure as Blackwell hardware arrives
Based on 2 verified sources

How to Negotiate CoreWeave Pricing

CoreWeave contracts are negotiable. These 5 tactics are sourced from real buyer experiences and procurement specialists.

Negotiation Playbook 5 tactics
Commit to High Volume for Discounted Rates high success

CoreWeave and similar neoclouds offer volume-based pricing for customers who commit to large, predictable compute volumes. Document your expected GPU hours over 12-24 months before entering negotiations. Third-party case studies show that AI companies have successfully negotiated discounted compute rates by committing to sustained traffic volumes.

Reddit (AI Dungeon/Latitude official post on negotiating discounted AI compute through volume commitments, May 2024)
Reserved Contracts vs On-Demand Pricing high success

Negotiate Reserved capacity contracts rather than paying on-demand rates. Reserved contracts lock in significantly lower per-hour GPU costs for baseline workloads. Use Spot instances for burst or fault-tolerant training jobs to further reduce costs beyond Reserved pricing.

Reddit (CoreWeave pricing tier structure discussions)
Bundle Baseline Reserved with Spot for Training high success

Structure contracts to use Reserved instances only for inference (steady-state load) and use Spot instances for training runs which can tolerate interruption. This hybrid approach minimizes reserved commitment size while capturing the lowest possible spot pricing for batch workloads.

Reddit (CoreWeave pricing structure and Spot tier discussions)
Use Competitor Quotes as Leverage medium success

CoreWeave prices at approximately 40-50% of AWS for similar instances, but competitors like Nebius position themselves as even cheaper. Obtain written quotes from Nebius, Lambda Labs, and Vast.ai before negotiating with CoreWeave. Neoclouds are in active competition for enterprise accounts and will often match or undercut competitor pricing.

Reddit (stocks subreddit, March 2025 — 'Just based on public pricing, Coreweave is about 40-50% of price of AWS for similar instances')
Negotiate Early for Blackwell Access medium success

CoreWeave has priority access to NVIDIA's latest GPU hardware including Blackwell (GB200). For teams that need cutting-edge hardware, committing to a CoreWeave contract early — before Blackwell availability broadens — may yield better allocation terms than waiting for open availability.

Reddit (NBIS_Stock discussions on CoreWeave-NVIDIA priority GPU access relationship, 2025)
High success Medium Low / difficult

Full negotiation guide →

CoreWeave Pricing FAQ

01 How much does CoreWeave cost?

CoreWeave GPU instances range from $10.00/hr (8x L40) to $68.80/hr (8x HGX B200) for on-demand pricing. Popular 8x H100 instances cost $49.24/hr. Spot pricing offers savings up to 54%. Reserved capacity with committed terms saves up to 60% over on-demand.

02 Does CoreWeave charge for data transfer?

No. CoreWeave does not charge for ingress, egress, or data transfer fees. Storage pricing ranges from $0.015/GB/month (cold) to $0.070/GB/month (distributed file storage).

03 What GPUs does CoreWeave offer?

CoreWeave offers NVIDIA GB300 NVL72, GB200 NVL72, HGX B200, RTX PRO 6000 Blackwell, HGX H200, HGX H100, GH200, L40S, L40, and A100 GPUs across North America and Europe regions.

04 What is CoreWeave's spot pricing?

CoreWeave's Spot instances are preemptible and offer up to 54% savings over on-demand rates. For example, the RTX PRO 6000 Blackwell is available at $9.24/hr on spot. Spot is best for fault-tolerant batch workloads that can handle interruptions.

05 CoreWeave vs Lambda: which is better for AI training?

CoreWeave focuses on enterprise-scale multi-GPU clusters and offers more GPU variety including the latest HGX B200 at $68.80/hr for 8 GPUs. Lambda offers simpler access and lower minimum commitments, with H100 SXM at $3.99/GPU/hr in 8-GPU configs. CoreWeave's strength is its enterprise SLAs and no data transfer fees.

06 How does CoreWeave pricing compare to AWS or Azure?

CoreWeave is priced at approximately 40-50% of AWS for comparable GPU instances. As a neocloud focused exclusively on AI GPU compute rather than a full cloud ecosystem, CoreWeave can offer lower rates. However, all CoreWeave pricing is custom-quoted through account managers — there are no published list prices for enterprise-scale workloads.

07 Does CoreWeave have a self-serve signup or free trial?

No. CoreWeave requires new customers to submit a contact form and go through a curated 'Request a Meeting' onboarding process. There is no self-serve account creation or free trial tier. This enterprise-focused model is designed for teams with established GPU compute requirements.

08 What is the difference between CoreWeave's On-Demand, Spot, and Reserved tiers?

CoreWeave offers three compute pricing models: On-Demand (single GPU or 8x GPU configurations, billed hourly with no long-term commitment), Spot (interruptible instances at lower rates, suitable for fault-tolerant training workloads), and Reserved (long-term contracted capacity at the lowest per-hour rates, standard for enterprise customers). All tiers are custom-priced based on GPU type, configuration, and scale.

09 Is CoreWeave suitable for small teams or individual developers?

CoreWeave is primarily designed for large enterprise AI teams. Its onboarding requires meeting with an account manager, and smaller users report receiving less attention as CoreWeave focuses on large-scale data center contracts. Smaller teams may find better self-serve options with Lambda Labs, RunPod, or Vast.ai.

10 Who are CoreWeave's major customers?

CoreWeave's confirmed major customers include Microsoft, Meta, IBM, and OpenAI. Microsoft is the dominant customer, representing a substantial portion of CoreWeave's revenue. This customer concentration is considered a risk factor for CoreWeave's business.

11 Does CoreWeave have a relationship with NVIDIA that affects GPU availability?

Yes. NVIDIA has historically given CoreWeave priority access to its latest GPU hardware, including early Blackwell (GB200) allocations. This is because CoreWeave, unlike hyperscalers, does not build competing AI chips. This relationship has helped CoreWeave secure GPU supply ahead of AWS, Azure, and Google Cloud.

Is this pricing incorrect? — we'll verify and update it.