Flat-rate, predictable, forecastable
Slices you can size and a bill you can forecast, instead of token shock and surprise egress.
Community
For evaluation and side projects.
- Core-VM trial credits
- Object Storage 5 GB
- Community support
- Full CLI access
Pro
Predictable flat rates for teams shipping in production.
- App Platform + Core-VM
- Flat-rate Neural inference
- Nodes & RPC endpoints
- Email support, 99.9% SLA
Enterprise
Sovereignty, scale and a 15-minute engineer SLA.
- Dedicated compute slices
- SecNumCloud-ready path
- DPA, subprocessors, residency
- 15-min engineer SLA, 24/7
See what predictable inference saves you
Compare per-token hyperscaler inference against a Citadea flat-rate slice.
Flat-rate tiers
Citadea Neural
Starter
€199/mo- 1 dedicated LPU
- Open-source models
- 10M tokens/month included
Pro
€499/mo- Dedicated LPUs
- Unlimited tokens*
- Latency SLA < 20 ms
Enterprise
€1999/mo- Dedicated compute slice
- 15-min engineer support
- SecNumCloud-ready path
Citadea Agent Hosting
Team
€499/mo- 5 agents
- SandBox included
- 10 GB Memory
Scale
€1499/mo- 50 agents
- MCP Gateway
- Multi-agent orchestration
* "Unlimited" reflects a dedicated slice sized to your throughput; fair-use applies. Prices are indicative pre-launch and exclude VAT.
Answers for engineers and buyers
What is Citadea?
Citadea is the sovereign developer cloud, a fortress for modern web, web3 and agent workloads, operated by Yggdrasil Digital SASU from Toulouse, France.
Are you exposed to the US Cloud Act?
No. Citadea is a 100% European company with no US parent, so it falls outside the reach of the US CLOUD Act, unlike US hyperscalers and their EU subsidiaries.
How does pricing work?
Predictable monthly flat rates, not per-token roulette. Use the token-shock calculator on /pricing to compare against hyperscaler inference.
What is "token shock"?
The unpredictable invoice you get when per-token AI pricing meets a multi-agent reasoning loop. Citadea Neural and Agent Hosting use flat rates to remove it.
Can I deploy a web app from git?
Yes, App Platform builds and deploys full-stack apps from a repository, with preview deployments and automatic HTTPS.
Which chains do Nodes & RPC support?
Major EVM and non-EVM chains, with archive data, WebSocket and trace/debug. Dedicated or shared endpoints, EU-hosted.
What is the support SLA?
15-minute tier-1 engineer response for Enterprise, 24/7/365, a systems engineer, not a script.
Is there a CLI?
Yes, one CLI drives everything. Install and quickstart commands are on /docs; public beta is coming soon.