Private AI Gateway.
Use Frontier Models Privately.
AI models such as GPT or Claude running in your own private environment in the region you choose.
What You Get.
One API, Many Models.
GPT 4.1 (5+ coming soon), Claude Opus & Sonnet, Gemini, Gemma, Mistral & others via one OpenAI-compatible endpoint. Switch models without switching providers.
Your Region. Your Rules.
Choose your region from EU (GDPR incl. UK, CH), CH, GER, UK, US, or AU. Your API key is bound to an isolated tenant in that region without cross-region routing.

Zero Retention, by Default.
We never log, store, or train on your data. A stateless AI gateway with no memory layer on our side and no shadow profiles. Model providers never get any data.
Works with the AI Tools You Already Use.
Chats
Open WebUI
Anything LLM
LobeChat
LibreChat
Odysseus- + more
Agents
OpenClaw
Hermes
Agent Zero
Langchain
LlamaIndex
CrewAI- + more
Coding
Claude Code
Cursor
Cline
Codex
GitHub Copilot
VS Code- + more
Automation & Workflows
n8n
LangFlow
Flowise- + more
One API. One Account.
Many Models, Private & Secure.
amazee.ai vs. ChatGPT, Claude & Gemini
| amazee.ai | OpenAI | Anthropic | ||
| Models | GPT, Claude, Gemini + Open Weights | GPT only | Claude only | Gemini only |
|---|---|---|---|---|
| Location | EU, CH, GER, UK, US, AU choice | US | US | US |
| Isolation | ✅ Isolated | ❌ | ❌ | ❌ |
| Audit reports | ✅ ISO 27001, SOC2 Type II | Enterprise only | Enterprise only | Enterprise only |
| Memory | Yours | Stored (US) | Stored (US) | Stored (US) |
| Pricing | Pay per Token | $20 | $20 | $20 |
import OpenAI from 'openai';
const openai = new OpenAI({
apiKey: process.env.AMAZEE_AI_API_KEY,
baseURL: 'https://llm.[REGION].amazee.ai/v1',
});
const chat = await openai.chat.completions.create({
model: 'openai/gpt-4.1', // Any model from our directory
messages: [{ role: 'user', content: 'Hello!' }],
});
from openai import OpenAI
import os
client = OpenAI(
api_key=os.environ["AMAZEE_AI_API_KEY"],
base_url="https://llm.[REGION].amazee.ai/v1",
)
chat = client.chat.completions.create(
model="openai/gpt-4.1", # Any model from our directory
messages=[{"role": "user", "content": "Hello!"}],
)
# Any model from our directory
curl https://llm.[REGION].amazee.ai/v1/chat/completions \
-H "Authorization: Bearer $AMAZEE_AI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "openai/gpt-4.1",
"messages": [{"role": "user", "content": "Hello!"}]
}'
Get Started in 5 Minutes.
Full self-service. Sign up, add payment, and generate API keys in your chosen region. Use it in almost any Bring-Your-Own-Key AI tool to work and scale securely and compliantly with Private AI.
Core
Self-serve.
- Manage multiple regions and API keys with spending limits for teams
- AI frontier models GPT 4.1 (5+ coming soon), Claude Opus & Sonnet, Gemini, Mistral, Gemma and more
- Isolated Tenants
- Regions available: EU (GDPR incl. UK, CH), CH, GER, UK, US, AU
- Full GDPR compliance, ISO 27001 · SOC 2 Type II
- Standard DPA
- Token-based pricing, no per-seat, no minimums
- Standard support (ticket queue)
Pay per token. Prices vary per model (see Model Directory). Credit card or voucher required.
Enterprise
Custom solutions such as dedicated infrastructure, enterprise support, or additional models.
- Dedicated cluster, your own isolated infrastructure
- Enterprise Data Processing Agreements (DPA)
- Models in additional regions
- Custom TPM / RPM, sized to your workload
- Fine-tuning: adapt LLMs to your domain or use case
- 99% uptime SLA with 2-hour response
- Priority Enterprise Support; direct line, not the ticket queue
- Infrastructure audits for your security team
- Custom contract terms, including payment schedules
Model Directory
Why amazee.ai is different
Other Gateways Might Read Your Input. We Don't.
Most AI gateways inspect, classify, or modify your content through middleware for PII scrubbing, prompt injection detection, or content-aware processing. We only route requests and never read your content. Our tenant-isolated infrastructure has zero data retention. Enterprises can deploy a dedicated AI gateway in our infrastructure or their own. Instead of managing multiple providers like Azure OpenAI or AWS Bedrock, you get access to leading model families with a single API key and DPA, while we handle model hosting and operations with our global engineering team.






















