# Fantasti

> GPU compute for AI teams, behind one API, one console and one bill.

The platform opens to its first cohort by invitation. No account is open yet. Until yours opens, nothing is provisioned and nothing is billed.

Fantasti builds the Fantasti Cloud Platform. Private Preview. Access opens in cohorts. Companies can request a place in the first, and places are limited. Designed to be operated end to end by AI agents, under written rules and hard limits. The desks open with the first cohort. Until then the owner reads every request and answers it.

- Request access: https://fantasti.ai/contact/
- The owner reads every request for the first cohort, and an approved request is invited by a sign-up link that is tied to the email address, valid for 72 hours and works once.
- Requests from personal email addresses join the waitlist for the next phase.
- API (Preview API): Not reachable yet. The address is published when the first cohort opens. One client for the whole platform: Python, REST and a CLI over the same objects, ids and limits. The SDK and the CLI are published when the first cohort opens.
- MCP server (Early access): Not reachable yet. The address is published when the first cohort opens. A remote MCP server that gives AI agents the Fantasti API as tools, inside the limits your admins set.
- Preview list prices. They apply when your account opens. Nothing is billed before that. In USD, last reviewed 2026-10-10. Metered by the second, billed hourly.

Every public page of fantasti.ai is listed below, one line per page: title, description, status.

Status words used on this site:

- Private Preview: Opens with the first cohort. Access by request.
- Early access: Being built for the first cohort. Join by request. Details may change.
- By request: Sold through sales on reserved terms. Not a release stage.
- Planned: On the roadmap, described as direction. No date.

## Products

- [GPU Instances](https://fantasti.ai/compute/#instances): Virtual machines with one or eight NVIDIA GPUs, on demand, on spot or reserved. Compute. Status: Private Preview.
- [CPU Instances](https://fantasti.ai/compute/#cpu-instances): General-purpose virtual machines from 2 to 256 vCPU with 4 GiB of memory per vCPU. Compute. Status: Private Preview.
- [GPU Clusters](https://fantasti.ai/clusters/): Groups of 8-GPU nodes joined by InfiniBand on a single fabric, delivered as an isolated Kubernetes cluster or as plain machines. Compute. Status: Private Preview.
- [Rack-scale](https://fantasti.ai/compute/#rack-scale): GB300 NVL72 and GB200 NVL72 racks, 72 GPUs in one NVLink domain, on reserved terms. Compute. Status: By request.
- [Spot](https://fantasti.ai/spot/): GPU capacity at a market price, with an optional max price per GPU-hour. Compute. Status: Private Preview.
- [Workspaces](https://fantasti.ai/workspaces/): GPU development environments that start on one GPU and scale the same Ray code to an autoscaling GPU cluster. Develop. Status: Early access.
- [Sandboxes](https://fantasti.ai/sandboxes/): Isolated, short-lived CPU and GPU environments that AI agents create and end through the API. Develop. Status: Early access.
- [Flows](https://fantasti.ai/flows/): Python workflows that run step by step on Fantasti GPUs, versioned and scheduled, from an experiment to production. Develop. Status: Planned.
- [Storage](https://fantasti.ai/storage/): Volumes, shared filesystems and S3-compatible object storage in the same region as your compute. Data and network. Status: Private Preview.
- [Networking](https://fantasti.ai/networking/): Private networks, public addresses, security groups, load balancers and private DNS, plus the InfiniBand fabric for clusters. Data and network. Status: Private Preview.
- [Batch Inference](https://fantasti.ai/batch-inference/): Offline inference over a dataset on spot or on-demand capacity, with checkpoints, retries and a max price. Run. Status: Private Preview.
- [Serverless](https://fantasti.ai/serverless/): Run a container on a GPU as a job that finishes or an endpoint that serves requests, with no machines to manage. Run. Status: Early access.
- [Integrations](https://fantasti.ai/integrations/): Run Kubernetes tools, Ray, SkyPilot or NVIDIA OSMO on your Fantasti cluster the way their docs describe. Run. Status: Early access.
- [API](https://fantasti.ai/api/): One client for the whole platform: Python, REST and a CLI over the same objects, ids and limits. Access. Status: Private Preview.
- [MCP server](https://fantasti.ai/mcp/): A remote MCP server that gives AI agents the Fantasti API as tools, inside the limits your admins set. Access. Status: Early access.
- [Console and one bill](https://fantasti.ai/pricing/#one-bill): One console and one statement across every product, split by cost center. Access. Status: Private Preview.
- [Orchestrator](https://fantasti.ai/orchestrator/): The control plane that checks policy, finds capacity and quota, places each workload on one fabric, works the spot market inside your max price, recovers work after a stop and meters it onto one statement. Operate. Status: Private Preview.
- [Observability](https://fantasti.ai/observability/): Metrics, logs and traces for everything on Fantasti, with prebuilt dashboards. Operate. Status: Early access.
- [MLflow](https://fantasti.ai/integrations/#mlflow): Managed MLflow tracking and model registry, wired into Workspaces, Flows, Batch Inference, Serverless jobs and Clusters. Operate. Status: Early access.
- [Applications](https://fantasti.ai/integrations/#applications): One-click applications installed on your cluster: notebooks, inference servers, vector databases, UIs. Operate. Status: Planned.

## Prices

- [NVIDIA B300](https://fantasti.ai/gpus/b300/): $10.45 per GPU-hour on demand. Spot: From $1.09. Reserved: request a quote.
- [NVIDIA B200](https://fantasti.ai/gpus/b200/): $9.35 per GPU-hour on demand. Spot: From $1.09. Reserved: request a quote.
- [NVIDIA H200 SXM](https://fantasti.ai/gpus/h200-sxm/): $5.94 per GPU-hour on demand. Spot: From $0.87. Reserved: request a quote.
- [NVIDIA H100 SXM5](https://fantasti.ai/gpus/h100-sxm5/): $5.40 per GPU-hour on demand. Spot: From $0.87. Reserved: request a quote.
- [NVIDIA RTX PRO 6000 Blackwell Server Edition](https://fantasti.ai/gpus/rtx-pro-6000/): $2.16 per GPU-hour on demand. Spot: From $0.87. Reserved: request a quote.
- [NVIDIA L40S](https://fantasti.ai/gpus/l40s/): $1.86 per GPU-hour on demand. List price for the base shape: 1 GPU, 8 vCPU, 32 GiB. Larger shapes are quoted. No spot. Reserved: request a quote.
- [NVIDIA GB300 NVL72](https://fantasti.ai/gpus/gb300-nvl72/): Reserved, contact sales.
- [NVIDIA GB200 NVL72](https://fantasti.ai/gpus/gb200-nvl72/): Reserved, contact sales.
- [CPU Instances](https://fantasti.ai/compute/#cpu-instances): $0.0396 per vCPU-hour. Includes 4 GiB of memory per vCPU. Shapes from 2 to 256 vCPU. On-demand capacity.
- [Volume, SSD](https://fantasti.ai/storage/#billing): $0.0888 per GiB-month. Metered on: provisioned size.
- [Volume, SSD, non-redundant](https://fantasti.ai/storage/#billing): $0.0663 per GiB-month. Metered on: provisioned size.
- [Volume, SSD, mirrored](https://fantasti.ai/storage/#billing): $0.1475 per GiB-month. Metered on: provisioned size.
- [Snapshot](https://fantasti.ai/storage/#billing): $0.0888 per GiB-month. Metered on: provisioned size.
- [Shared filesystem](https://fantasti.ai/storage/#billing): $0.1000 per GiB-month. Metered on: provisioned size.
- [Bucket, Standard](https://fantasti.ai/storage/#billing): $0.0184 per GiB-month. Metered on: stored size.
- [Bucket, Tiered, warm tier](https://fantasti.ai/storage/#billing): $0.0375 per GiB-month. Metered on: stored size.
- [Bucket, Tiered, cold tier](https://fantasti.ai/storage/#billing): $0.0125 per GiB-month. Metered on: stored size.
- [Bucket, High throughput](https://fantasti.ai/storage/#billing): $0.1375 per GiB-month. Metered on: stored size.
- [Standard class requests: PUT, POST, DELETE](https://fantasti.ai/storage/#billing): $4.25 per million requests. Metered on: requests.
- [Standard class requests: GET, HEAD](https://fantasti.ai/storage/#billing): $0.35 per million requests. Metered on: requests.
- [Standard class egress, to the internet or another region](https://fantasti.ai/storage/#billing): $0.0188 per GiB. Metered on: data transferred.
- [Tiered class egress, above three times the stored volume](https://fantasti.ai/storage/#billing): $0.0625 per GiB. Metered on: data transferred.
- [Support plan, Basic](https://fantasti.ai/support/#plans): Included with every account. For: everyone.
- [Support plan, Developer](https://fantasti.ai/support/#plans): $29 per month. For: individual developers.
- [Support plan, Team](https://fantasti.ai/support/#plans): $100 per month. For: startups running production.
- [Support plan, Direct](https://fantasti.ai/support/#plans): $1,000 per month. For: teams with reserved or multi-node capacity.
- [Support plan, Enterprise](https://fantasti.ai/support/#plans): Custom. For: large organisations.

"From" is the lowest spot price for that GPU. The spot price moves with supply and demand.
Reserved rates are quoted per term.

## Home

- [Fantasti · GPU cloud, from one GPU to rack scale](https://fantasti.ai/): The Fantasti Cloud Platform: NVIDIA GPUs, CPU machines, clusters and storage behind one API and one bill, placed by the Fantasti Orchestrator and metered by the second. Status: Private Preview.

## Platform

- [Fantasti Cloud Platform: every product on one account](https://fantasti.ai/platform/): Compute, development environments, storage, networking and operations on one API, one set of policies and one statement. How the Fantasti Cloud Platform works. Status: Private Preview.
- [Compute: GPU and CPU instances, and rack-scale systems](https://fantasti.ai/compute/): Virtual machines with one or eight NVIDIA GPUs, CPU machines from 2 to 256 vCPU, and GB300 NVL72 and GB200 NVL72 racks on reserved terms. Metered by the second. Status: Private Preview.
- [GPU Clusters: 8-GPU nodes on one InfiniBand fabric](https://fantasti.ai/clusters/): Multi-node NVIDIA GPU clusters on a single InfiniBand fabric, as an isolated Kubernetes cluster or plain machines. On-demand, spot or reserved. Status: Private Preview.
- [Spot: market-priced GPUs with a max price](https://fantasti.ai/spot/): Follow the spot price or set the most you will pay per GPU-hour. You pay the spot price, never your max. A stop notice before nodes stop, and disks are kept. Status: Private Preview.
- [Storage: volumes, shared filesystems and S3-compatible buckets](https://fantasti.ai/storage/): Block, file and object storage in the same region as your GPUs. Shared filesystems up to 100 PiB, three object storage classes, one bill. Status: Private Preview.
- [Networking: private networks, security groups and InfiniBand](https://fantasti.ai/networking/): Private networks, public addresses, stateful security groups, load balancers and private DNS, with an InfiniBand fabric between cluster GPUs. Status: Private Preview.
- [Observability: GPU metrics, logs and traces](https://fantasti.ai/observability/): GPU, InfiniBand, node and storage metrics collected automatically. PromQL, LogQL and OpenTelemetry, with prebuilt dashboards. Status: Early access.
- [Flows: Python workflows from experiment to production](https://fantasti.ai/flows/): Write ML workflows as Python flows and run each step on the GPU it needs. Versioned runs, schedules and events, evals as gates, spot with a max price, one bill. Status: Planned.
- [Orchestrator: policy, placement, the Broker and quota](https://fantasti.ai/orchestrator/): The Fantasti Orchestrator checks every request against your policy, places it on one fabric, works the spot market inside your max price and grows your quota as you grow. Status: Private Preview.
- [Agents: how Fantasti is built to operate](https://fantasti.ai/agents/): Fantasti is designed to be operated end to end by AI agents with written job descriptions, decision limits and hard limits. High-stakes decisions pass a second, senior review, and every decision is logged. Status: Early access.
- [MCP server: the Fantasti API for AI agents](https://fantasti.ai/mcp/): A remote MCP server for Cursor, VS Code, Codex and other MCP clients such as coding agents. Sign in with OAuth or a scoped agent key. Agents act inside the quotas, spend caps and price ceilings your admins set. Status: Early access.
- [Workspaces: GPU dev environments that scale to Ray clusters](https://fantasti.ai/workspaces/): Develop in VS Code or JupyterLab on one GPU, then run the same Ray code on an autoscaling GPU cluster. Spot workers with a max price, files that persist, one bill. Status: Early access.
- [Sandboxes: short-lived GPU environments for AI agents](https://fantasti.ai/sandboxes/): Isolated CPU and GPU sandboxes your agents create and end through the API, for RL, evals, GPU CI and untrusted code. Time-to-live, network modes, per-second metering. Status: Early access.
- [Batch Inference: offline inference at market prices](https://fantasti.ai/batch-inference/): Offline inference and evaluation over a dataset on spot or on-demand GPUs, with checkpoints, retries and a max price. Status: Private Preview.
- [Serverless: GPU jobs and endpoints without clusters](https://fantasti.ai/serverless/): Run a container on a GPU as a job that finishes or an endpoint that serves requests. Metered per second, spot with a max price, one bill. Status: Early access.
- [Integrations: Kubernetes, Ray, SkyPilot and NVIDIA OSMO](https://fantasti.ai/integrations/): Run the scheduler you already use on an isolated Fantasti cluster with its own Kubernetes API. Managed MLflow and one-click applications. Status: Early access.

## GPUs

- [GPU catalog: NVIDIA Blackwell, Hopper and Ada](https://fantasti.ai/gpus/): Specifications and preview list prices for B300, B200, RTX PRO 6000, H200, H100 and L40S, and for GB300 NVL72 and GB200 NVL72 racks. Reviewed 2026-10-10.
- [NVIDIA B300](https://fantasti.ai/gpus/b300/): Specifications, node shapes and Fantasti listing status for NVIDIA B300. NVIDIA reference figures, reviewed 2026-10-10. Status: Private Preview.
- [NVIDIA B200](https://fantasti.ai/gpus/b200/): Specifications, node shapes and Fantasti listing status for NVIDIA B200. NVIDIA reference figures, reviewed 2026-10-10. Status: Private Preview.
- [NVIDIA H200 SXM](https://fantasti.ai/gpus/h200-sxm/): Specifications, node shapes and Fantasti listing status for NVIDIA H200 SXM. NVIDIA reference figures, reviewed 2026-10-10. Status: Private Preview.
- [NVIDIA H100 SXM5](https://fantasti.ai/gpus/h100-sxm5/): Specifications, node shapes and Fantasti listing status for NVIDIA H100 SXM5. NVIDIA reference figures, reviewed 2026-10-10. Status: Private Preview.
- [NVIDIA RTX PRO 6000 Blackwell Server Edition](https://fantasti.ai/gpus/rtx-pro-6000/): Specifications, node shapes and Fantasti listing status for NVIDIA RTX PRO 6000 Blackwell Server Edition. NVIDIA reference figures, reviewed 2026-10-10. Status: Private Preview.
- [NVIDIA L40S](https://fantasti.ai/gpus/l40s/): Specifications, node shapes and Fantasti listing status for NVIDIA L40S. NVIDIA reference figures, reviewed 2026-10-10. Status: Private Preview.
- [NVIDIA GB300 NVL72](https://fantasti.ai/gpus/gb300-nvl72/): Specifications, node shapes and Fantasti listing status for NVIDIA GB300 NVL72. NVIDIA reference figures, reviewed 2026-10-10. Status: By request.
- [NVIDIA GB200 NVL72](https://fantasti.ai/gpus/gb200-nvl72/): Specifications, node shapes and Fantasti listing status for NVIDIA GB200 NVL72. NVIDIA reference figures, reviewed 2026-10-10. Status: By request.

## Pricing

- [Pricing: on-demand, spot and reserved GPU prices](https://fantasti.ai/pricing/): Preview list prices per GPU-hour, spot with a max price, reservations, three ways to pay and the spend threshold. Metered by the second, one statement. Reviewed 2026-10-10.

## Resources

- [API: one client for the Fantasti Cloud Platform](https://fantasti.ai/api/): Python, REST, a CLI and an MCP server over the same objects, ids and limits. Preview API, subject to change. Status: Preview API.
- [Support: plans, response targets and Slack](https://fantasti.ai/support/): Five support plans, from included to $1,000 per month. Set up so that AI agents take every case at any hour and say they are AI. Escalations go to senior review in the business days your plan states. Status: Early access.
- [Community: the forum for Fantasti members](https://fantasti.ai/community/): A members' forum for people who run workloads on Fantasti. Sign in with your console account. Fantasti agents take part, labelled as AI. Status: Planned.
- [Feedback: propose and vote on what Fantasti builds](https://fantasti.ai/feedback/): Members propose changes to Fantasti and vote on them. Every idea has a status, and every status change comes with a written answer. Status: Planned.
- [Previews: use Fantasti features before release](https://fantasti.ai/previews/): Preview programs give your team new Fantasti features while they are being built. Each one states its stage, its billing, its support and what may still change. Status: Early access.
- [Changelog: every change to the Fantasti Cloud Platform](https://fantasti.ai/changelog/): Dated entries for every change to the Fantasti Cloud Platform, by product and release stage. RSS and JSON feeds. Updated 2026-10-10.

## Company

- [Research](https://fantasti.ai/research/): Fantasti research in progress: network-based GPU virtualization, delivery protocols for GPU workloads, and placement and interruption prediction. Status: Research.
- [Trust and security](https://fantasti.ai/trust/): What you secure, what Fantasti secures and what the infrastructure layer secures. Every control has a status, and each infrastructure provider is listed with what it holds before a workload runs on it. Reviewed 2026-10-10.
- [Company](https://fantasti.ai/company/): Fantasti builds the Fantasti Cloud Platform: GPU compute for AI teams, designed to be placed, priced and supported by AI agents under written rules. What we are building, and why.
- [Request access](https://fantasti.ai/contact/): Request access to the Fantasti Cloud Platform. Companies can request a place in the first cohort, and no account is open yet. Tell us what you want to run.

## Legal

- [Legal](https://fantasti.ai/legal/): Terms of Service, Privacy Policy, Acceptable Use Policy and Service Level Agreement for Fantasti Technologies LLC.
- [Terms of Service](https://fantasti.ai/legal/terms/): Terms of Service for the Fantasti Cloud Platform.
- [Privacy Policy](https://fantasti.ai/legal/privacy/): How Fantasti Technologies LLC collects, uses and protects information.
- [Acceptable Use Policy](https://fantasti.ai/legal/aup/): What may not be run on Fantasti.
- [Service Level Agreement](https://fantasti.ai/legal/sla/): Service commitments and service credits.
- [Cluster requirements v1](https://fantasti.ai/reserved/requirements/v1/): What every Fantasti reserved cluster commits to before billing starts: fabric, tenancy, acceptance, replacement, support and disclosure.

## Feeds

- https://fantasti.ai/changelog/index.json
- https://fantasti.ai/changelog/rss.xml