Bluefin Router

by Naventic AI

One OpenAI-compatible API for AI models running on our own GPU servers.
Private beta · invited members

This address is for apps and code

Use the base URL below in your code or in any app that supports an "OpenAI-compatible" endpoint (Open WebUI, LibreChat, n8n, LangChain, Continue…), together with your personal API key. A members' web dashboard with sign-in, usage and a chat playground is coming soon.

Base URL:  https://bluefin.naventic.ai/v1
API key:   sk-…   (sent to you by email)

Checking service…

Quick test

curl https://bluefin.naventic.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"model":"bluefin-chat","messages":[{"role":"user","content":"Hello!"}]}'

Python

from openai import OpenAI
client = OpenAI(base_url="https://bluefin.naventic.ai/v1", api_key="YOUR_KEY")
r = client.chat.completions.create(model="bluefin-chat",
      messages=[{"role": "user", "content": "Hello!"}])
print(r.choices[0].message.content)

Models

bluefin-chat	Qwen3.8 Flash-Next: general chat, writing, coding, Bahasa Indonesia
bluefin-reasoning	GLM-5.3 Flash: harder reasoning, step-by-step answers
bluefin-fast	Hermes-4-14B: quick, short answers
bluefin-chat-lite	lighter Qwen3.8 Flash-Next: cheap, quick drafts
bluefin-embed	embeddings for search / RAG (1024 dimensions)

Most models serve one request at a time per machine, so a busy model may queue briefly. Each member starts with 1,000,000 tokens of credit. We store token counts only, never your prompts or replies.