Skip to main content
πŸ“– The AI Tool Bible
CoreWeave preview image
CoreWeave logo

CoreWeave

βœ“ Text source-checked Oct 24, 2026

An AI cloud platform purpose-built to turn complex AI workloads into discovery and creation.

Source-checked against the vendor's site Β· rebuilt from the vendor's own pagesProfile updated Link checked Pricing captured How we rate

Enterprise(not source-checked)Β· Free tier Β· from $0.015/moFine-tuningDeepSeek (not source-checked)8.2 / 10

In short

CoreWeave is an AI cloud platform for training and serving models on NVIDIA GPUs. It offers Kubernetes-native environments with automated cluster management and on-demand pricing.

CoreWeave supports training and serving on NVIDIA Vera Rubin NVL72, Blackwell, and Hopper architectures on bare metal within a Kubernetes-native environment.

CoreWeave Mission Control automates node lifecycle, health checks, and observability to maintain cluster goodput. CoreWeave offers Kubernetes Service and SUNK (Slurm on Kubernetes) for workload scheduling and cluster utilization.

CoreWeave offers on-demand GPU instances for workloads that do not require long-term capacity commitments. Pricing covers compute, storage, networking, reserved capacity, and managed Kubernetes offerings.

From the vendor's pages

According to CoreWeave: CoreWeave supports training and serving on NVIDIA Vera Rubin NVL72, Blackwell, and Hopper architectures on bare metal within a Kubernetes-native environment. Its pages list this strength: Supports the newest NVIDIA architectures including Vera Rubin NVL72, Blackwell, and Hopper on bare metal.

β€” The AI Tool Bible editorial team

Pros

  • βœ… Supports the newest NVIDIA architectures including Vera Rubin NVL72, Blackwell, and Hopper on bare metal.
  • βœ… Includes Mission Control for automated node lifecycle, health checks, and observability.
  • βœ… Provides Kubernetes Service and SUNK for Slurm-on-Kubernetes scheduling.
  • βœ… Offers ARENA to run real workloads on purpose-built infrastructure before production commitment.
  • βœ… Features NVIDIA Quantum InfiniBand and Spectrum-X Ethernet networking for cluster scale-out.

Pricing (captured )

TierMonthlyKey features
CoreWeave AI Object Storage Hot$0.06/GB/moβ€”
CoreWeave AI Object Storage Warm$0.03/GB/moβ€”
CoreWeave AI Object Storage Cold$0.015/GB/moβ€”
Distributed File Storage$0.070/GB/moβ€”
Egress and Input Output Operations per SecondFreeβ€”
Public IP Address$4.00 per IP / monthβ€”
Bring Your Own IP$4.00 one-time setup fee per IPβ€”

Prices change; we show the grid as captured on Nov 8, 2026. Check current pricing on the vendor's site β†’

Facts

Platforms
web (not source-checked)
Pricing
enterprise - Free tier Β· from $0.015/mo (not source-checked)
API available
Yes (not source-checked)
Company
CoreWeave, Inc. (not source-checked)
Model used
DeepSeek (not source-checked)
Hacker News
9 stories Β· top 4 points Β· last Sep 29, 2026 Β· checked Nov 10, 2026
Earliest Wayback capture
2019-07-17 (earliest archived capture of the site, not a launch date)
Founded
2017 (Wikidata)
Headquarters
United States (Wikidata)

Where this tool sits

Use cases

AI model trainingAI inferenceVFX and renderingReinforcement learning

Frequently asked

How much does CoreWeave cost for on-demand GPU instances?
On-demand pricing varies by hardware: NVIDIA HGX H100 is $49.24/hour, HGX B200 is $68.80/hour, GB200 NVL72 is $42.00/hour, and A100 is $21.60/hour. These rates cover compute, storage, networking, and managed Kubernetes offerings.
What hardware architectures does CoreWeave support for AI workloads?
CoreWeave supports training and serving on NVIDIA Vera Rubin NVL72, Blackwell, and Hopper architectures. These run on bare metal within a Kubernetes-native environment, suitable for AI model training, inference, and reinforcement learning.
How does CoreWeave handle cluster management and scheduling?
Mission Control automates node lifecycle, health checks, and observability to maintain cluster goodput. For scheduling, CoreWeave offers Kubernetes Service and SUNK (Slurm on Kubernetes) to optimize cluster utilization for various AI workloads.
Is CoreWeave suitable for workloads without long-term commitments?
Yes, CoreWeave offers on-demand GPU instances specifically for workloads that do not require long-term capacity commitments. This allows users to scale resources as needed for tasks like VFX, rendering, or model training.
What specific AI use cases is CoreWeave designed for?
CoreWeave is purpose-built for AI model training, AI inference, VFX and rendering, and reinforcement learning. It turns complex AI workloads into discovery and creation through its specialized cloud platform.

Compare CoreWeave with similar tools

All alternatives β†’

Explore related

Reviews