Pricing
A small monthly base for your dedicated instance. Then per-minute usage on the compute each agent needs, from €0.009 an hour to H100 GPUs. Idle agents cost nothing.
One dedicated instance for a small team getting its first departments into production.
Elastic worker pools, GPU agents and governance for a company running across departments.
A dedicated control plane in the region you choose, with private networking and contractual terms.
Usage
| Profile | vCPU | Memory | GPU | Good for | Plans | Per hour |
|---|---|---|---|---|---|---|
| Standard | 2 | 4 GB | — | Most agents: chat, research, writing, back office, CRM work. | Launch, Scale | €0.009 |
| Performance | 4 | 8 GB | — | Coding agents, builds, browser automation, heavier tool use. | Launch, Scale | €0.039 |
| High memory | 4 | 32 GB | — | Large datasets, analytics, knowledge-graph consolidation. | Scale | €0.14 |
| GPU · L4 | 8 | 32 GB | 1× NVIDIA L4 24 GB | Local inference, embeddings, transcription, image work. | Scale | €0.59 |
| GPU · A100 | 12 | 85 GB | 1× NVIDIA A100 80 GB | Fine-tuning, large open models, video pipelines. | Scale | €2.90 |
| GPU · H100 | 16 | 120 GB | 1× NVIDIA H100 80 GB | The heaviest training and inference workloads. | Scale | €4.90 |
Set the exact resources an agent needs.
Agent workspaces, files, database and backups. Launch includes 50 GB, Scale 500 GB.
€0.12 per extra GB-month
Bring your own Anthropic, OpenAI or Google keys and pay the provider directly with no markup, or use the managed model gateway billed at provider list price plus 5%.
Estimate
| Compute profile | Agents | Running h/day each | Month |
|---|---|---|---|
Standard €0.009/h | €5.40 | ||
Performance €0.039/h | €0 | ||
High memory €0.14/h | Scale plan | ||
GPU · L4 €0.59/h | Scale plan | ||
GPU · A100 €2.90/h | Scale plan | ||
GPU · H100 €4.90/h | Scale plan |
Estimate over 30 days, excluding VAT and model tokens. Billed per minute of running time; idle agents are suspended and not billed.
Start on Launch| Launch | Scale | Enterprise | ||
|---|---|---|---|---|
| Where it runs | ExoHuman Cloud | ExoHuman Cloud | ExoHuman Cloud | |
| Seats | 5 | Unlimited | Unlimited | |
| Agents | 10 | Unlimited | Unlimited | |
| Concurrent agent runs | 3 | 50 | Sized to you | |
| GPU agents | — | Yes | Yes | |
| Backup retention | 7 days | 30 days | 90 days | |
| Single sign-on | — | OIDC | OIDC, SAML, SCIM | OIDC, SAML, SCIM |
| Uptime SLA | — | 99.9% | 99.95% | Per contract |
A monthly base fee for your dedicated instance (control plane, Workspace, backups, upgrades and monitoring), plus the running time of your agents on each compute profile, metered per minute. Storage above your plan's allowance is billed per GB-month.
No. An agent is billed only while its container runs a task. When it goes idle it is suspended; its memory and files stay in git and on your instance's storage, and it wakes on the next message, schedule or event.
Bring your own Anthropic, OpenAI or Google keys and pay the provider directly with no markup, or use the managed model gateway billed at provider list price plus 5%.
Yes. Upgrade from Launch to Scale at any time from your account; the base fee is prorated. Compute profiles can be enabled per agent from your instance's settings.
Never. Every customer gets a dedicated instance with its own database, credentials and agent containers. Worker capacity is isolated per instance.
Payments are processed by Stripe. Invoices, payment methods and tax IDs are managed in the billing portal linked from your account. Prices exclude VAT, which is added where applicable.
Yes, at any time from the billing portal. Your instance stays available until the end of the period, and you can export all agent repositories and data for 30 days afterwards.
Yes. Connect any Linux, Windows or macOS server, VM or workstation as a worker node in two steps. Agents you place there run on your hardware — capacity, GPUs and network stay inside your perimeter, with the same approvals and audit trail.