Fluidstack

aid: fluidstack
name: Fluidstack
description: >-
  Fluidstack is an AI cloud platform that builds and operates high-performance,
  single-tenant GPU clusters for top AI labs, governments, and enterprises.
  Founded in 2017 out of Oxford University and now headquartered in New York
  City, Fluidstack manages more than 100,000 GPUs across its global network and
  has been selected by customers including Anthropic, Meta, Mistral, Poolside,
  Black Forest Labs, and Character.AI. Its core offering combines on-demand
  bare-metal GPU clusters (including thousands of NVIDIA H200s with InfiniBand)
  with managed Kubernetes and managed Slurm orchestration layers, Lighthouse
  observability, and human-on-call site reliability. Fluidstack exposes a
  global REST Management API (api.atlas.fluidstack.io/api/v1alpha1) and the
  fluidctl CLI for managing organizations, projects, regions, clusters,
  instances, members, invitations, and SSH keys. The platform is GDPR, SOC 2
  Type 2, and ISO 27001 certified, and was awarded ClusterMAX Gold by
  SemiAnalysis.
type: Index
position: Provider
access: 3rd-Party
image: https://kinlane-productions.s3.amazonaws.com/apis-json/apis-json-logo.jpg
tags:
  - AI
  - Artificial Intelligence
  - GPU
  - Cloud
  - Compute
  - Infrastructure
  - Machine Learning
  - Foundation Models
  - Training
  - Inference
  - Kubernetes
  - Slurm
  - Bare Metal
  - NVIDIA
  - InfiniBand
  - Data Centers
url: https://raw.githubusercontent.com/api-evangelist/fluidstack/refs/heads/main/apis.yml
created: '2026-05-25'
modified: '2026-05-25'
specificationVersion: '0.20'
apis:
  - aid: fluidstack:fluidstack-management-api
    name: Fluidstack Management API
    description: >-
      The global Fluidstack Management API (v1alpha1) for managing GPU cloud
      resources. Exposes regions, clusters, organizations, members,
      invitations, instances, and user SSH keys. Uses bearer token
      authentication, optional X-PROJECT-ID header scoping for instance
      listings, and is served from https://api.atlas.fluidstack.io/api/v1alpha1.
    humanURL: https://docs.fluidstack.io/api-reference/management-api
    baseURL: https://api.atlas.fluidstack.io/api/v1alpha1
    tags:
      - AI
      - GPU
      - Cloud
      - Compute
      - Clusters
      - Instances
      - Management
    properties:
      - url: https://docs.fluidstack.io/api-reference/management-api
        type: Documentation
      - url: openapi/fluidstack-management-api-openapi.yml
        type: OpenAPI
      - url: https://docs.fluidstack.io/redocusaurus/management-api.yaml
        type: OpenAPI
common:
  - type: Website
    url: https://www.fluidstack.io
  - type: Portal
    url: https://docs.fluidstack.io
  - type: Documentation
    url: https://docs.fluidstack.io/getting-started/overview
  - type: GettingStarted
    url: https://docs.fluidstack.io/getting-started/overview
  - type: Documentation
    url: https://docs.fluidstack.io/projects/overview
  - type: Documentation
    url: https://docs.fluidstack.io/kubernetes/overview
  - type: Documentation
    url: https://docs.fluidstack.io/slurm/overview
  - type: Documentation
    url: https://docs.fluidstack.io/lighthouse/overview
  - type: Documentation
    url: https://docs.fluidstack.io/api-reference/management-api
  - type: SDK
    url: https://docs.fluidstack.io/cli-reference/fluidctl
    name: fluidctl CLI
  - type: About
    url: https://www.fluidstack.io/about-us/about
  - type: Blog
    url: https://www.fluidstack.io/about-us/blog
  - type: Careers
    url: https://www.fluidstack.io/about-us/careers
  - type: Contact
    url: https://www.fluidstack.io/contact
  - type: LinkedIn
    url: https://www.linkedin.com/company/fluidstack
  - type: Twitter
    url: https://twitter.com/fluidstackio
  - type: Features
    data:
      - On-demand single-tenant GPU clusters with isolated hardware, network, and storage
      - Immediate access to thousands of NVIDIA H200 GPUs with InfiniBand interconnect
      - Managed Kubernetes — bare-metal container orchestration with NVIDIA operators and health checks
      - Managed Slurm — bare-metal batch orchestration with user management and topology-aware scheduling
      - Lighthouse observability and monitoring for GPU clusters
      - Global Management API (v1alpha1) at api.atlas.fluidstack.io for clusters, instances, regions, members, invitations, and SSH keys
      - fluidctl command-line tool with auth, capacity, instance-types, instances, kubernetes, projects, slurm commands
      - Bearer token authentication with optional X-PROJECT-ID header scoping
      - Custom data centers — selected by Anthropic to deliver capacity in New York and Texas
      - Demonstrated rapid deployment — 2,500+ GPUs delivered to Poolside in 48 hours
      - 100,000+ GPUs under management across the network
      - 15-minute response SLA with direct engineer maintenance and audit logs
      - GDPR, SOC 2 Type 2, and ISO 27001 certified
      - ClusterMAX Gold awarded by SemiAnalysis (Nov 2025)
    sources:
      - https://www.fluidstack.io
      - https://docs.fluidstack.io
      - https://docs.fluidstack.io/api-reference/management-api
      - https://docs.fluidstack.io/cli-reference/fluidctl
      - https://www.fluidstack.io/about-us/blog
maintainers:
  - FN: Kin Lane
    email: [email protected]
    X: apievangelist
    url: https://apievangelist.com