# CoreWeave (slim)

> CoreWeave is a GPU cloud that rents NVIDIA GPU nodes through a managed Kubernetes service (CKS), with REST and gRPC platform APIs, a Terraform provider and a hosted MCP server for observability.

- Full: https://www.anchorterminal.com/tools/coreweave.md (~8,050 tokens) · this version ~1,630 tokens · JSON https://www.anchorterminal.com/tools/coreweave.json · canonical https://www.anchorterminal.com/tools/coreweave
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · Updated: 2026-10-08

**C · 61.5/100 · rank #363 of 722 · #5 in GPU & serverless compute · not agent-ready · confidence medium**


Assessment: Per-hour prices for 8-GPU H100, H200 and B200 nodes are public, the docs ship as Markdown with llms.txt and embedded OpenAPI specs, and IAM has read-only roles per service. An organisation must be approved by CoreWeave's sales team before any token exists, and no request rate limits were found in the reviewed documentation.

## Facts

- Kind: HTTP API · vendor: CoreWeave, Inc. · category: GPU & serverless compute · legal entity: CoreWeave, Inc. · provenance 83/100
- Endpoint: `https://api.coreweave.com` (HTTP)
- Auth: Token · pricing: Pay per use · x402: no · licence: Proprietary service under CoreWeave's Terms of Service. The Terraform provider on GitHub is MIT
- Probe metrics: not measured yet (probes haven't run)
- Free tier: None found. Trials run through the sales-led ARENA programme
- Access: Sales approves the organisation, then tokens are created in the Cloud Console
- GPUs: GB300 and GB200 NVL72, HGX B300, B200, H200 and H100, A100, L40S, L40, GH200 and RTX PRO 6000 Blackwell, mostly as 8-GPU nodes. Vera Rubin NVL72 in limited availability from 30 September 2026
- Scale to zero: Node Pools take `targetNodes`, `minNodes` and `maxNodes` with an optional autoscaler. Nodes bill while they exist
- Billing basis: Per instance-hour, monthly in arrears. Invoices under $20,000 are charged to the card on file
- Rate limits: No request rate limits found. The gateway caps Kubernetes API requests at about 50 KB of URL and headers and 100 MB of body
- APIs: CKS (v1beta1), VPC, Object Storage, Dedicated Inference (v1alpha1), Capacity Finder, Telemetry Relay and FOCUS usage export at api.coreweave.com, REST and gRPC, with generated clients on the Buf Schema Registry
- MCP: Mission Control at https://mc.coreweave.com/mcp, streamable HTTP, 38 documented tools, 2 of them write tools enabled per organisation
- Service level: Objective of 99.9 per cent across multiple regions and 99 per cent in one, with credits of up to 50 per cent of the month's bill
- Regions: Seven super regions (US-EAST, US-CENTRAL, US-WEST, CA-EAST, EU-NORTH, EU-SOUTH, EU-WEST)
- Prices: HGX H100 80 GB $6.16 per GPU-hour; HGX H200 141 GB $6.31 per GPU-hour; HGX B200 180 GB $8.60 per GPU-hour; GB200 NVL72 186 GB $10.50 per GPU-hour; A100 80 GB $2.70 per GPU-hour; L40S 48 GB $2.25 per GPU-hour; AI Object Storage, hot tier $0.06 per GB per month

- Scores: Reliability 50, Performance pending, Schema & documentation 79, Agent ergonomics 58, Security & auth 74, Payments & pricing 20, Task success pending, Maintenance & community 80, Transparency & trust 77 · total over the 7 assessed categories
- Why: Reliability, Graded as a hosted service on the CKS and platform APIs. · Schema & documentation, OpenAPI 3.0 specs for the CKS, VPC, Object Storage, Dedicated Inference, Capacity Finder and Telemetry Relay APIs are embedded in the refere… · Agent ergonomics, Graded on the platform API, the Kubernetes API and the Mission Control MCP server together. · Security & auth, API access tokens are user-scoped, carry an expiry, are shown once and can be revoked in the console. · Payments & pricing, No machine payment protocol (0). · Maintenance & community, The changelog's latest entry is 2 October 2026 (30). · Transparency & trust, Closed service with published terms.
- Sources: 26, open questions: 7, both in the full twin
- Capabilities: compute.gpu, compute.containers, compute.endpoints, compute.batch
- JSON: https://www.anchorterminal.com/api/v1/tools/coreweave.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/coreweave.svg` or a link to https://www.anchorterminal.com/tools/coreweave from a page on coreweave.com or one of its subdomains, or the README of github.com/coreweave/terraform-provider-coreweave, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.

## Before you call it

1. Create an API access token at console.coreweave.com/tokens with an expiry and send it as `Authorization: Bearer` to https://api.coreweave.com
2. Add GPUs by applying a NodePool resource (`compute.coreweave.com/v1alpha1`) to a CKS cluster with `instanceType` and `targetNodes`. `instanceType` can't be changed afterwards
3. Set `targetNodes` to 0 or delete the Node Pool when a job ends. Nodes are whole 8-GPU machines billed by the hour
4. Never pass `dry_run` to the MCP tool `coreweave_kubectl_apply` expecting a preview. The reference says the parameter is ignored and the manifest is applied
5. Use a separate Forge API key and https://api.inference.wandb.ai/v1 for Serverless Inference. A CoreWeave API access token doesn't work there

## Connect

```bash
curl "https://api.coreweave.com/v1beta1/cks/clusters" -H "Authorization: Bearer $COREWEAVE_API_TOKEN"
```

```bash
claude mcp add --transport http mission-control https://mc.coreweave.com/mcp --header "Authorization: Bearer [YOUR-COREWEAVE-API-TOKEN]"
```
Full config and headless snippets are in the full page. Through letme (picks today, calling later): https://letme.dev/coreweave

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| Modal | B | 63.6 | compute.gpu, compute.endpoints, compute.batch, compute.containers | https://www.anchorterminal.com/tools/modal.min.md |
| Northflank | C | 61.3 | compute.gpu, compute.containers, compute.batch, compute.endpoints | https://www.anchorterminal.com/tools/northflank.min.md |
| Beam | C | 55.5 | compute.gpu, compute.endpoints, compute.batch, compute.containers | https://www.anchorterminal.com/tools/beam.min.md |
| Cerebrium | C | 55.3 | compute.gpu, compute.endpoints, compute.batch, compute.containers | https://www.anchorterminal.com/tools/cerebrium.min.md |
| Runpod | D | 53.5 | compute.gpu, compute.endpoints, compute.batch, compute.containers | https://www.anchorterminal.com/tools/runpod.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)