Live — Altores

The AI platform your
data never leaves

An OpenAI-compatible API running entirely on your own infrastructure. LLM inference, document OCR, metered usage, and a full developer dashboard — with zero data leaving your environment.

Book a demo Deployment options
OpenAI-compatible API
0 bytesleave your infrastructure
GDPRMonaco-law compliant

Core Capabilities

An enterprise AI API, entirely under your control

SAIP provides the full stack required to build and run AI-powered systems in sovereignty-constrained environments — LLM inference, document processing, developer tooling, and operational oversight.

LLM API — OpenAI-compatible

The same API surface as OpenAI — POST /v1/chat/completions and GET /v1/models — but running on GPU hardware you own. Streaming responses, multi-turn conversations, and multimodal inputs are all supported.

Drop-in replacement for the OpenAI SDK
Streaming responses via server-sent events
Multimodal support (text + image inputs)
Configurable per-key rate limits

Deployment

Run anywhere. Own everything.

SAIP is a self-contained infrastructure component. It runs on a single GPU server, a sovereign cloud cluster, or in a fully air-gapped data centre with no external network dependencies.

Managed

Hosted by Altores on Monaco Cloud. Zero infrastructure management. Full data sovereignty within Monaco jurisdiction.

Sovereign Cloud

Deployed on your sovereign cloud tenancy. You own the infrastructure; Altores handles deployment and updates.

On-Premise

Runs on your GPU hardware in your data centre. Full control over hardware, network, and data residency.

Air-Gapped

Deployed in an environment with no external network access. Docker images delivered offline. No internet dependency.

Hardware requirement

SAIP runs on a single NVIDIA GPU server (minimum 24 GB VRAM for LLM + OCR). Altores can advise on hardware configuration for your usage profile.

Deployment options →

Use Cases

Built for organisations where data sovereignty is not optional

Telecom

AI on subscriber data, in-network

Run LLM inference on customer records, call transcripts, and support history without any data leaving your network. SAIP powers the AI layer of the Support Portal and CRM without third-party API calls.

Zero subscriber data leaves the network
Powers Altores Intelligence in-platform
OpenAI SDK drop-in for existing apps
Public Services

Government-grade AI infrastructure

Meet national AI sovereignty mandates and data localisation requirements. SAIP can be deployed in a government data centre or sovereign cloud with full air-gap support and no foreign data transfer.

Air-gapped deployment available
Meets national data localisation rules
Audit log for every AI interaction
Finance & Legal

AI for confidential documents

Process legal contracts, financial statements, and compliance reports with the OCR API — entirely within your own infrastructure. No document content ever transmitted to a third-party cloud.

On-premise OCR pipeline
No document data leaves the environment
Exportable audit trail for regulators

Frequently Asked Questions

SAIP runs entirely within your own infrastructure — your cloud, your data centre, or an air-gapped environment. No data is ever sent to third-party AI providers. The API is fully OpenAI-compatible, so existing integrations work without modification, but all inference happens on hardware you control.
Yes. SAIP exposes an OpenAI-compatible REST API (/v1/chat/completions, /v1/models). Any application currently using the OpenAI SDK can switch to SAIP by updating the base URL and API key — no code changes required.
SAIP comes with a curated model catalogue including Altores's own production models, instruction-tuned LLMs for chat and reasoning, and specialised models for document OCR. The model catalogue is configurable per deployment — Altores can onboard specific models for your use case, including models you supply.
Yes. The OCR API (POST /v1/ocr) accepts PDF, PNG, and JPG inputs and returns rich Markdown output including tables, lists, and headings. Jobs are processed asynchronously — you submit a file, get a job ID, and poll for the result. It's purpose-built for high-quality extraction from complex documents.
SAIP includes a full metering and billing layer. API usage is tracked per client with configurable monthly quotas, overage rules, and pay-as-you-go options. Usage data is surfaced in the dashboard with per-model breakdowns. Billing integration is available for organisations requiring automated invoicing.
Yes. SAIP is deployable in fully air-gapped environments with no external network access. Docker images can be exported as tarballs for offline delivery. All inference runs locally on your GPU hardware. The deployment tooling supports this mode explicitly.

Deploy your sovereign AI platform

We handle the full deployment — hardware sizing, installation, model onboarding, API configuration, and integration with the Altores platform. Your team gets a production-ready AI API in days.

Book a demo Deployment options