Platform

Products

Private AI infrastructure — from compute to agents.

Compute

GPU Instances
Dedicated GPUs, billed hourly
BYOK & Managed Keys
OpenAI, Anthropic, custom

Intelligence

Agents & Smart Routing
Virtual models, cost control
Fine-Tuning (LoRA)
Your data, your models
Knowledge Bases & RAG
Documents, vector search
Semantic Cache
Lower costs with AI caching

Integration

MCP Tools
Tools, APIs, external services
API Gateway
OpenAI-compatible endpoint
Data sovereignty
On-prem, EU hosting, GDPR

Target groups

Use Cases

For enterprise, SMBs, and individual developers.

Enterprise

Compliance & Privacy
GDPR, no US cloud provider
Team rollout
Workspace, roles, API gateway
Cost Control
Smart routing reduces token costs

SMB

Quick start with managed keys
No GPU required, start immediately
Your own assistants
Chatbot with RAG + system prompt
Fine-tuning with domain knowledge
Small model, precise answers

Developers & Individuals

OpenClaw / OpenCode
Private and on-prem agents
MCP Integration
Tools directly in model context
OpenAI compatibility
Drop-in replacement, no SDK switch

Knowledge & Support

Resources

Everything you need to succeed with Mycelis.

Learn

Documentation
API reference & guides
Learn
Tutorials & video courses
Quickstart
Be productive in 5 minutes

Community & Updates

Blog
News, guides & deep dives
Changelog
What is new in Mycelis?

Support

Status
Uptime & incidents
Contact
hello@mycelis.io

Integration

API Gateway —
OpenAI-compatible endpoint for everything.

Drop-in replacement for the OpenAI API. Change only base_url and api_key — no additional code changes required.

Start for free API Reference

2-line migration from OpenAI

Before (OpenAI direct)

client = OpenAI(
  api_key="sk-..."
)

After (Mycelis Gateway)

client = OpenAI(
  base_url="https://api.mycelis.io/proxy/v1",
  api_key="pat_..."
)

Supported endpoints

Endpoint Function Status
POST /chat/completions Chat, streaming, function calling ✓
POST /embeddings Text embeddings ✓
GET /models List available models ✓
POST /completions Legacy completions (OpenAI v1) ✓
POST /images/generations Image generation (DALL-E) Soon

Gateway features

PAT authentication

Personal access tokens with configurable scopes — read-only, write, admin.

Usage logging

Complete request logging: timestamp, model, tokens, latency, costs.

Rate limiting

Configurable rate limits per PAT, workspace, or model.

GDPR-compliant proxy

No logging of prompt content. EU data centers, DPA available on request.

Frequently asked questions

Is the API Gateway fully OpenAI-compatible?

Yes. Implemented: /chat/completions (including streaming and function calling), /embeddings, /models, /completions. All official OpenAI SDKs work without code changes.

How do I generate an API key (PAT)?

In the Mycelis dashboard under 'API Keys'. The token is shown once, then only stored as a hashed fingerprint. You can create and revoke PATs individually.

Can I use the API Gateway with Cursor, ChatGPT Desktop, or other tools?

Yes. Any tool that supports an OpenAI-compatible endpoint can be used with Mycelis. Set base_url to https://api.mycelis.io/proxy/v1 and api_key to your PAT.

Are there SLA guarantees?

The API Gateway has 99.9% uptime (30-day average). For enterprise SLAs, contact sales@mycelis.io.

Drop-in replacement for OpenAI — in 2 minutes.

Your existing code keeps running. Only change base_url and api_key.

Start for free