Auxen
Wire your coding agent to Auxen so it can provision dedicated-GPU private endpoints for Llama, Qwen, or Mistral billed per minute.
Overview
ai.auxen/auxen is a MCP server for the Build phase that lets agents provision and use Auxen private GPU model endpoints (Llama, Qwen, Mistral) over Streamable HTTP.
What is this MCP server?
- Remote MCP at https://api.auxen.ai/mcp (Streamable HTTP)
- Dedicated GPU private endpoints for Llama, Qwen, and Mistral families
- Pay-per-minute pricing model via Auxen dashboard
- OAuth 2.1 + PKCE for browser clients or Bearer auxen_live_* / auxen_test_* API keys
- RFC 8414 / 9728 discovery and optional Dynamic Client Registration for connectors directories
- 1 production MCP URL: https://api.auxen.ai/mcp
- 2 key prefixes: auxen_live_* and auxen_test_*
- 3 model families called out: Llama, Qwen, Mistral
What problem does it solve?
Running private LLMs on dedicated GPUs from an agent usually means custom dashboards, keys, and glue outside your MCP workflow.
Who is it for?
Solo builders shipping agents or backends that need dedicated-GPU hosting and want minute-based billing from the same chat or IDE tooling.
Skip if: Teams that only need cheap shared API tokens with no private GPUs, or anyone avoiding cloud OAuth and API key setup.
What do I get? / Deliverables
Your MCP client can talk to Auxen’s hosted control plane with OAuth or API keys and stand up pay-per-minute private inference endpoints.
- MCP-authenticated access to Auxen private model endpoints
- OAuth or API-key based connection to https://api.auxen.ai/mcp
- Runnable dedicated-GPU stacks for supported open model families
Recommended MCP Servers
Journey fit
Private inference endpoints are built-phase integration work when your product or agent needs models outside shared APIs. Integrations is the right shelf because this MCP exposes Auxen’s hosted control plane over Streamable HTTP rather than shipping inference code locally.
How it compares
GPU endpoint provisioning MCP, not a local Ollama skill or a single static model wrapper.
Common Questions / FAQ
Who is ai.auxen/auxen for?
Developers using MCP clients who want Auxen-managed private models on dedicated GPUs with dashboard billing and key or OAuth auth.
When should I use ai.auxen/auxen?
Use it during build when you integrate agents with private Llama, Qwen, or Mistral endpoints instead of only public shared APIs.
How do I add ai.auxen/auxen to my agent?
Register https://api.auxen.ai/mcp as a remote Streamable HTTP server, then complete OAuth 2.1 + PKCE or set Authorization Bearer to an auxen_live_* or auxen_test_* key from auxen.ai/dashboard/api-keys.