Your own security model, deployed in one click
A deployment is a private instance of a security model on a dedicated, EU-hosted GPU. Your tools talk to it through an OpenAI-compatible API, and your data never leaves it.
Why a dedicated deployment
Security work involves client data, unpublished findings and offensive technique. A deployment of your own gives you control over all three.
Your data stays with you
Scan output, configs, samples and findings are processed on your own instance. Inference traffic goes straight to your endpoint.
A GPU that is only yours
Each deployment hosts one model on a dedicated GPU. Nothing is shared with other customers, and there are no shared rate limits.
The model you chose, unchanged
The instance runs exactly the version you deployed. Results stay reproducible, and updates never arrive without notice.
Built for security work
The weights are adapted for cybersecurity tasks by specialised developers, in a Blue Team and a Red Team variant where the licence allows it. The model engages with the work you actually do.
Costs you can plan
A flat hourly price while the instance runs. No token metering, so heavy triage costs the same as a quiet hour.
EU hosting, documented
EU GPU providers certified to ISO/IEC 27001, a company with a certified ISMS, and a data processing agreement as part of onboarding.
Your data never passes through our control plane
Requests travel encrypted from your tools straight to your dedicated GPU. Our control plane provisions the instance, but it is never in the data path.
Your side
- Terminal, scripts
- Agents, coding harness
- Chat UI, IDE, SIEM
Your dedicated instance · EU
Your own GPU cluster
- One model
- One customer
- Nothing is stored
Your data never passes through InferPod.
InferPod
Platform
Deploys and manages your instance.
- ISO/IEC 27001 certified company ISMS
- EU GPU providers only
- Providers certified to ISO/IEC 27001
What you get
Every deployment is the same simple stack, from your private endpoint down to the GPU. All of it belongs to your instance.
- Private endpointOpenAI-compatible API over TLS, only reachable with your key
- API key checkEach instance has its own key, checked on the instance
- Security modelOpen-weight model with weights adapted for cybersecurity
- Dedicated GPUYour own hardware with an EU GPU provider
Works with the tools you already use
Security frameworks, coding agents and your own code: point them at your instance with a base URL, an API key and a model name.
Agentic security tools and frameworks run their own recon, testing and analysis loops against your private model.
export STRIX_LLM="openai/your-security-model"
export LLM_API_KEY="YOUR_API_KEY"
export LLM_API_BASE="https://your-instance/v1"
strix --target ./your-appStrix reaches its model through LiteLLM, so a custom base URL is all it needs.
Illustrative. Any tool, agent or platform that supports a custom OpenAI-compatible endpoint works the same way: base URL, API key, model name.
How teams use a deployment
Short-lived for a single project, or running all day for a team.
Per engagement
Start an instance when a project begins, keep client artefacts inside it, and delete it when the engagement ends.
Behind your agents
Point a coding agent, an agent harness or your own security platform at a private endpoint, without changing the workflow.
Always on for a team
Keep an instance running for your SOC or research group, with a pinned model version and a predictable hourly cost.
Ready to deploy?
Select a model, choose a location and call your API.
Select a model
Choose your variant
Deploy
Choose a location
Call your API
Endpoint and key, ready to use
URL https://de-a1b2c3.example.com/v1 KEY sk-••••••••••••••••- VariantBlue Team
- LocationGermany
Illustrative. Works with any OpenAI-compatible client.
Deployment questions
What is a deployment?
A deployment is your own running instance: one model on one dedicated GPU, with a private OpenAI-compatible endpoint and its own API key.
Do I have to manage servers?
No. You choose a model, a variant and a location, and the GPU is provisioned for you. There is no infrastructure to manage.
Which tools work with it?
Any client that speaks the OpenAI API: SDKs, coding agents, agent harnesses and agentic security platforms. You change the base URL and the key.
Who can reach my endpoint?
Only callers with the API key of your instance, over TLS. IP allowlisting and private networking are planned for later.
What happens to my data when I stop the instance?
Stopping deletes the instance. Instance content is not kept afterwards, and prompts and outputs are never stored while it runs.
Where can I deploy?
In the EU, at launch in France, Germany, Poland and Finland, with GPU providers that are certified to ISO/IEC 27001. Your instance always runs in the country you choose.
Request early access
Tell us about your team and the models you want to run. We will contact you as soon as your tier is available.