Mycelis
Products Model hosting BYOK & Managed Keys Agents & Smart Routing Fine-Tuning (LoRA) Knowledge Bases & RAG MCP Tools API Gateway
Use Cases Compliance & Privacy Team rollout Cost Control Quick start with managed keys Your own assistants Fine-tuning with domain knowledge OpenClaw / OpenCode MCP Integration OpenAI compatibility
Resources Getting Started Documentation Blog Status
Models MCP Hub Pricing FAQ

API gateway

One endpoint.Every model in your workspace.

Your applications, editors and agents call a single base URL. Mycelis routes each request to the right model — hosted on our GPUs or a commercial provider.

No base fee · No minimum term · Pause any time

What the gateway does

  • OpenAI-compatibleSame request and response format, streaming included
  • Routing rulesPer request, by tokens, keywords or tool use
  • Personal-data filterKeep sensitive requests off external providers
  • Usage trackingWhich model answered and what it cost

What you get

The API you already use — with more control.

Change the base URL and the key. Everything else in your setup stays as it is.

Drop-in compatible

Existing OpenAI SDKs, editors and agent frameworks keep working with the Mycelis base URL.

Access keys

Personal access tokens per user or application, so access is easy to grant and revoke.

Routing

Send each request to the model your rules choose, or let the automatic router decide.

Personal-data filter

Detect personal data in requests and keep such requests on your own hosted models.

Traceable decisions

Every request records which model answered — no black box.

Usage and costs

See consumption per workspace and model.

How it works

Connect in three steps.

01

Create an access key

Generate a key in your workspace.

02

Set the base URL

Point your SDK or editor at https://api.mycelis.ai/proxy/v1.

03

Call any activated model

Use the model name from your workspace — hosted or commercial.

Business model

No fee for the gateway itself.

You pay for the components behind it, and only for what you use.

Hosted models

Billed by runtime at the GPU rate while they run.

Commercial models

Billed per token — with your key or a managed key.

No platform fee

No seats, no base fee, no minimum term.

cURL
curl https://api.mycelis.ai/proxy/v1/chat/completions \
  -H "Authorization: Bearer pat_..." \
  -H "Content-Type: application/json" \
  -d '{
        "model": "your-model-name",
        "messages": [{"role": "user", "content": "Hello"}]
      }'

Common questions

Questions about the gateway.

In most projects you change two things: the base URL and the API key.

Yes. Request and response formats, including streaming, follow the OpenAI API.

Yes. Both are reachable through the same endpoint, and routing rules can send requests to either.

Any client that lets you set an OpenAI-compatible base URL — for example Cursor, Continue, Claude Code or your own SDK calls.

Ready to get started?

One base URL. Every model.

Create a free account and point your first application at Mycelis.