Quickstart

Muna serves open models like Qwen 3.8 27B and Gemma 4 26B behind OpenAI- and Anthropic-compatible APIs. If a tool lets you set a base URL and an API key, it works with Muna.

API	Base URL
OpenAI	`https://inference.muna.ai/v1`
Anthropic	`https://inference.muna.ai`

Get an API Key

export MUNA_API_KEY="<your Muna API key>"

$env:MUNA_API_KEY = "<your Muna API key>"

Keep your API key on your machine or your server. Do not ship it in browser or mobile app code.

Use a Coding Agent

Point your agent at Muna, then start it in your project directory:

Claude Code
OpenCode
Pi
Hermes

export ANTHROPIC_BASE_URL="https://inference.muna.ai"
export ANTHROPIC_AUTH_TOKEN="$MUNA_API_KEY"
export ANTHROPIC_MODEL="@qwen/qwen-3.8-27b"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="@qwen/qwen-3.8-27b"

claude

See Claude Code for details.

Add Muna as a provider in opencode.json:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "muna": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Muna",
      "options": {
        "baseURL": "https://inference.muna.ai/v1",
        "apiKey": "{env:MUNA_API_KEY}"
      },
      "models": {
        "@qwen/qwen-3.8-27b": { "name": "Qwen 3.8 27B" }
      }
    }
  }
}

See OpenCode for details.

Add Muna to ~/.pi/agent/models.json:

{
  "providers": {
    "muna": {
      "baseUrl": "https://inference.muna.ai/v1",
      "api": "openai-completions",
      "apiKey": "$MUNA_API_KEY",
      "models": [{ "id": "@qwen/qwen-3.8-27b" }]
    }
  }
}

See Pi for details.

Set Muna as a custom endpoint in ~/.hermes/config.yaml:

model:
  default: "@qwen/qwen-3.8-27b"
  provider: custom
  base_url: https://inference.muna.ai/v1
  key_env: MUNA_API_KEY

See Hermes for details.

Then try:

Explain how this repository is organized.

Build an Application

Use the official OpenAI or Anthropic SDK. Only the base URL and the API key change:

from openai import OpenAI
import os

client = OpenAI(
    base_url="https://inference.muna.ai/v1",
    api_key=os.environ["MUNA_API_KEY"]
)

completion = client.chat.completions.create(
    model="@qwen/qwen-3.8-27b",
    messages=[{ "role": "user", "content": "Say hello in one sentence." }]
)
print(completion.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://inference.muna.ai/v1",
  apiKey: process.env.MUNA_API_KEY
});

const completion = await client.chat.completions.create({
  model: "@qwen/qwen-3.8-27b",
  messages: [{ role: "user", content: "Say hello in one sentence." }]
});
console.log(completion.choices[0].message.content);

from anthropic import Anthropic
import os

client = Anthropic(
    base_url="https://inference.muna.ai",
    api_key=os.environ["MUNA_API_KEY"]
)

message = client.messages.create(
    model="@qwen/qwen-3.8-27b",
    max_tokens=1024,
    messages=[{ "role": "user", "content": "Say hello in one sentence." }]
)
print(message.content[0].text)

curl https://inference.muna.ai/v1/chat/completions \
  -H "Authorization: Bearer $MUNA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "@qwen/qwen-3.8-27b",
    "messages": [{ "role": "user", "content": "Say hello in one sentence." }]
  }'

Next Steps

Models

Every live model, with input, cached input, and output prices.

OpenAI API

Chat completions, streaming, embeddings, and what is supported.

Anthropic API

The Messages API for Claude-compatible tools and SDKs.

Get Started

Coding Agents

Local Inference

Get an API Key

Use a Coding Agent

Build an Application

Next Steps

Models

OpenAI API

Anthropic API

​Get an API Key

​Use a Coding Agent

​Build an Application

​Next Steps

Models

OpenAI API

Anthropic API

Get an API Key

Use a Coding Agent

Build an Application

Next Steps