# Crusoe Cloud (slim)

> Crusoe Cloud rents NVIDIA and AMD GPU virtual machines with managed Kubernetes and Slurm, and runs hosted model inference, dedicated deployments and LoRA fine-tuning. Resources are managed through a REST API, a CLI, a Terraform provider and a Go client.

- Full: https://www.anchorterminal.com/tools/crusoe-cloud.md (~9,200 tokens) · this version ~2,330 tokens · JSON https://www.anchorterminal.com/tools/crusoe-cloud.json · canonical https://www.anchorterminal.com/tools/crusoe-cloud
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · Updated: 2026-10-10

**B · 63.4/100 · rank #396 of 950 · #6 in GPU & serverless compute · not agent-ready · confidence medium**


Assessment: The v1 REST API has a published OpenAPI description, Markdown docs, HMAC-signed keys, reader roles and a 90-day audit log, with on-demand GPU prices public. No idempotency keys or control-plane rate limits were found, spot and B200 prices go through sales, and a storage outage in eu-norway1 lasted about four hours on 29 September 2026.

## Facts

- Kind: HTTP API · vendor: Crusoe · category: GPU & serverless compute · legal entity: Crusoe Technologies LLC · provenance 90/100
- Endpoint: `https://api.cloud.crusoe.ai/v1` (HTTP, stdio)
- Auth: API key · pricing: Pay per use · x402: no · licence: Proprietary service under the Crusoe Cloud Platform Terms of Service. The Go client and the MCP server package are MIT. The CLI ships as binaries, and no licence file was found in the Terraform provider repository
- Probe metrics: not measured yet (probes haven't run)
- Surface graded: The REST API at `https://api.cloud.crusoe.ai/v1` (Swagger 2.0). The CLI, the Terraform provider and the preview MCP server are noted alongside
- MCP server: Preview, local, stdio, `npx -y @crusoeai/cloud-mcp` (0.1.6, MIT). 41 read-only tools named in the docs across discovery, compute, storage, networking, Kubernetes, audit logs, quotas and usage. Lists are cut to 50 items by default
- Free tier: $5 of introductory credits for Serverless Inference and Serverless Fine-Tuning with no payment method. None found for GPU machines
- GPUs: NVIDIA GB200, B300, B200, H200, H100, A100 and L40S, AMD MI355X and MI300X. L40S and A100 PCIe come in 1 to 10 GPU shapes, the others as 8-GPU or 4-GPU machines (https://docs.crusoecloud.com/compute/virtual-machines/overview)
- Orchestration: Crusoe Managed Kubernetes at $0.10 a cluster-hour, Managed Slurm, a container registry and instance templates with bulk create
- Scale to zero: None found. Self-Serve Deployments run a set replica count and bill until deleted. An on-demand VM can be stopped and is then billed only for its disks
- Cold start: No figure published. A VM in the `Creating` state can take up to a minute per the docs
- Billing basis: On-demand VMs per second in the running state. Reserved VMs for the whole commitment. Self-Serve Deployments per GPU-hour, Serverless Inference and Fine-Tuning per token. Ingress and egress are free
- Rate limits: None found for the infrastructure API. Serverless Inference allows 500,000 tokens and 30 requests a minute per model without a payment method, 2,000,000 and 600 with one, and returns `x-ratelimit-*` headers
- Errors: HTTP 400, 401, 403, 404 and 500 with a `code` and `message` body. No error catalogue found
- Pagination: `limit` with `next_token` and `prev_token`, or `page_size` with `next_page_token`, on 16 list calls. VM lists filter by id, name, type, location and state
- Credentials: Access key ID and secret key created in the console, with optional alias and expiry, deletable, inheriting the creator's roles. Separate Intelligence API keys for Managed AI. Service accounts with their own credentials appear in the API description and the Terraform provider 1.6.0
- Roles: `org-admin`, `org-editor`, `org-reader`, `project-editor` and `project-reader`. SSO through OIDC, MFA and SCIM
- Audit: 90 days of control-plane actions at `/organizations/{org_id}/audit-logs`, admins only, no read or list events (https://docs.crusoecloud.com/identity-and-security/audit-logs)
- SLA: 99.5 per cent monthly uptime per GPU virtual machine and per persistent disk volume, with credits of 10, 25 or 100 per cent, claimed within 30 days. Version 8.0, effective 21 July 2026 (https://legal.crusoe.ai/#service-level-agreements)
- Regions: Seven data centre zones in Texas, Ohio, Nevada, Virginia, Iceland (two) and Norway (https://docs.crusoecloud.com/locations)
- SDKs: Go `github.com/crusoecloud/client-go` v1.0.18 (1 October 2026, MIT, generated from the API description), Terraform provider 1.6.0 changelog entry of 8 October 2026, and a closed-source CLI
- Certifications: ISO 27001, ISO 42001 and SOC 2 Type I and II per https://www.crusoe.ai/cloud. The trust centre at trust.crusoe.ai is drawn by script and was not read
- Status: status.crusoecloud.com on Atlassian Statuspage, with components for GPU machines in six regions, networking, storage, Kubernetes, API, console, container registry and Intelligence Foundry
- Prices: NVIDIA H100 80 GB HGX, on demand $3.90 per GPU-hour; NVIDIA H200 141 GB HGX, on demand $4.29 per GPU-hour; NVIDIA A100 80 GB SXM, on demand $2.30 per GPU-hour; NVIDIA A100 80 GB PCIe, on demand $2 per GPU-hour; NVIDIA L40S 48 GB, on demand $1.50 per GPU-hour; AMD MI300X 192 GB, on demand $3.45 per GPU-hour; NVIDIA H100, Self-Serve Deployment $5.50 per GPU-hour; General-purpose CPU instance $0.04 per vCPU-hour; Persistent disk $0.08 per GB per month

- Scores: Reliability 66, Performance pending, Schema & documentation 79, Agent ergonomics 45, Security & auth 74, Payments & pricing 30, Task success pending, Maintenance & community 76, Transparency & trust 77 · total over the 7 assessed categories
- Why: Reliability, Graded as a hosted service on the REST API. · Schema & documentation, A Swagger 2.0 description for `/v1` and `/v1alpha5`, linked from the reference page and kept in `crusoecloud/client-go`, and an OpenAPI desc… · Agent ergonomics, Lists take `limit` or `page_size` on 16 of about 104 list calls, and VM lists filter by id, name, type, location and state; no field selecti… · Security & auth, Access key and secret key with optional expiry, deletable in the console or through `/users/tokens`. · Payments & pricing, No machine payment protocol (0). · Maintenance & community, Crusoe Updates entries on 5, 7 and 8 October 2026 and a Terraform provider 1.6.0 changelog entry on 8 October (30). · Transparency & trust, Closed service under Terms of Service version 1.10, effective 10 August 2026, with ten earlier versions kept on the legal page; the Go clien…
- Sources: 38, open questions: 13, both in the full twin
- Capabilities: compute.gpu, compute.containers, compute.batch, compute.endpoints
- JSON: https://www.anchorterminal.com/api/v1/tools/crusoe-cloud.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/crusoe-cloud.svg` or a link to https://www.anchorterminal.com/tools/crusoe-cloud from a page on crusoe.ai or one of its subdomains, or the README of github.com/crusoecloud/client-go, 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. Sign every request. Build the payload from path, sorted query string, verb and `X-Crusoe-Timestamp`, HMAC-SHA256 it with the base64url-decoded secret, and send `Authorization: Bearer 1.0:<access_key_id>:<signature>`.
2. Use `https://api.cloud.crusoe.ai/v1`. Some docs pages and the MCP server still name `api.crusoecloud.com` and `/v1alpha5`.
3. Creates, updates and deletes return an operation. Poll the matching operations endpoint for `state`, and list existing resources by name before retrying a create, because no idempotency key was found.
4. Managed AI uses separate Intelligence API keys and hosts, `api.inference.crusoecloud.com` for OpenAI-compatible calls and `api.intelligence.crusoecloud.com` for files and fine-tuning jobs.
5. Delete a VM and its disks to stop charges. A stopped on-demand VM is not billed for compute, but its 128 GB OS disk still bills at $0.08 a GiB-month.

## Connect

```bash
brew install crusoecloud/cli/crusoe
```

```bash
curl https://api.cloud.crusoe.ai/v1/capacities -H "X-Crusoe-Timestamp: <RFC3339 timestamp>" -H "Authorization: Bearer <version:access_key_id:base64_encoded_signature>"
```

```bash
claude mcp add crusoe-cloud -- npx -y @crusoeai/cloud-mcp
```
Full config and headless snippets are in the full page. Through letme (picks today, calling later): https://letme.dev/crusoe-cloud

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| Nebius AI Cloud | B | 67.2 | compute.gpu, compute.endpoints, compute.batch, compute.containers | https://www.anchorterminal.com/tools/nebius-ai-cloud.min.md |
| Modal | B | 63.6 | compute.gpu, compute.endpoints, compute.batch, compute.containers | https://www.anchorterminal.com/tools/modal.min.md |
| Verda | B | 62.3 | compute.gpu, compute.endpoints, compute.batch, compute.containers | https://www.anchorterminal.com/tools/verda.min.md |
| CoreWeave | C | 61.5 | compute.gpu, compute.containers, compute.endpoints, compute.batch | https://www.anchorterminal.com/tools/coreweave.min.md |
| Northflank | C | 61.3 | compute.gpu, compute.containers, compute.batch, compute.endpoints | https://www.anchorterminal.com/tools/northflank.min.md |


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