> ## Documentation Index
> Fetch the complete documentation index at: https://docs.emby.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Anthropic API Compatibility

> Use Emby’s Anthropic-compatible /v1/messages endpoint to run any Emby model through Claude-style APIs — fully compatible with Claude Code and Anthropic SDKs.

# Anthropic-Compatible API (Claude-Compatible)

Emby provides a native **Anthropic-compatible Messages API** at:

```
POST https://api.emby.dev/v1/messages
```

This allows you to use **any Emby model** — Kimi, DeepSeek, Qwen, GLM, Llama, and more — through the familiar **Claude-style request format**.

Perfect for:

* **Claude Code CLI**
* **Tools expecting Anthropic’s `/v1/messages` schema**
* **Apps that cannot easily switch from Anthropic → OpenAI format**
* **Teams wanting better performance + predictable costs + EU hosting**

***

# Why Use Emby’s Anthropic Endpoint?

<CardGroup cols={2}>
  <Card title="Drop-In Claude Compatibility" icon="code">
    Keep your existing Claude Code, scripts, and integrations — just point them at Emby.
  </Card>

  <Card title="Use Any Model" icon="model">
    Run Kimi, DeepSeek, Qwen, GLM, and more through the Anthropic API schema.
  </Card>

  <Card title="EU Hosting & Privacy" icon="shield">
    All Emby traffic stays inside the EU via bit.nl and Nebius Amsterdam.
  </Card>

  <Card title="Predictable Costs" icon="tag">
    No more unpredictable Anthropic token billing — just simple per-developer plans.
  </Card>
</CardGroup>

***

# Quick Setup (Claude Code)

Claude Code works instantly with Emby using three environment variables.

```bash theme={null}
export ANTHROPIC_BASE_URL="https://api.emby.dev/v1"
export ANTHROPIC_AUTH_TOKEN="<your-emby-key>"
# Optional: choose a model — otherwise Emby defaults to DeepSeek
export ANTHROPIC_MODEL="deepseek-v3"
```

Run Claude:

```bash theme={null}
claude
```

Your Claude Code CLI now uses **Emby models** with **Claude’s API format**.

***

# Choosing a Model

Use any Emby-hosted or routed model with Claude Code:

```bash theme={null}
# Fastest / most popular
export ANTHROPIC_MODEL=deepseek-v3

# Best reasoning (Open Source)
export ANTHROPIC_MODEL=kimi-k2-thinking

# Qwen Coder (for dev workflows)
export ANTHROPIC_MODEL=qwen2.5-coder-32b

# Ultra-cheap high-speed
export ANTHROPIC_MODEL=glm4-9b

# Claude (when using BYOK routing)
export ANTHROPIC_MODEL=anthropic/claude-3.5-sonnet
```

Any model Emby exposes is compatible.

***

# Environment Variables Explained

### `ANTHROPIC_MODEL`

Primary model your Claude-compatible tool uses.

```
export ANTHROPIC_MODEL=kimi-k2-thinking
```

### `ANTHROPIC_SMALL_FAST_MODEL`

Some Anthropic tools require a "fast fallback model."

```
export ANTHROPIC_SMALL_FAST_MODEL=glm4-9b
```

### Full Example

```bash theme={null}
export ANTHROPIC_BASE_URL="https://api.emby.dev/v1"
export ANTHROPIC_AUTH_TOKEN="<your-emby-key>"
export ANTHROPIC_MODEL="deepseek-v3"
export ANTHROPIC_SMALL_FAST_MODEL="glm4-9b"
```

***

# Manual API Request (Anthropic Style)

```bash theme={null}
curl -X POST "https://api.emby.dev/v1/messages" \
  -H "Authorization: Bearer $EMBY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v3",
    "messages": [
      {"role": "user", "content": "Explain Rust lifetimes simply."}
    ],
    "max_tokens": 150
  }'
```

***

# Response Format (Claude-Compatible)

Emby returns messages in **Anthropic’s native JSON structure**:

```json theme={null}
{
  "id": "msg_123",
  "type": "message",
  "role": "assistant",
  "model": "deepseek-v3",
  "content": [
    {
      "type": "text",
      "text": "Rust lifetimes ensure references remain valid…"
    }
  ],
  "stop_reason": "end_turn",
  "usage": {
    "input_tokens": 18,
    "output_tokens": 42
  }
}
```

You can swap models without changing any client code.

***

# Advanced Usage

## Switching Models Quickly

```bash theme={null}
export ANTHROPIC_MODEL=kimi-k2-thinking    # complex reasoning
export ANTHROPIC_MODEL=glm4-9b             # cheap & fast
export ANTHROPIC_MODEL=qwen2.5-coder-32b   # coding workflows
export ANTHROPIC_MODEL=deepseek-v3         # balanced default
```

## Persistent Configuration

Add to `.bashrc` or `.zshrc`:

```bash theme={null}
echo 'export ANTHROPIC_BASE_URL="https://api.emby.dev/v1"' >> ~/.zshrc
echo 'export ANTHROPIC_AUTH_TOKEN="<your-emby-key>"' >> ~/.zshrc
echo 'export ANTHROPIC_MODEL="deepseek-v3"' >> ~/.zshrc
source ~/.zshrc
```

***

# Need Help?

We help teams migrate their Claude workflows in minutes.

<CardGroup cols={2}>
  <Card title="WhatsApp Support" icon="phone">
    Chat with us instantly:<br />
    <a href="https://wa.absolum.nl" target="_blank">[https://wa.absolum.nl](https://wa.absolum.nl)</a>
  </Card>

  <Card title="Book a Call" icon="calendar">
    For enterprise routing, BYOK, or custom GPU servers:<br />
    <a href="https://cal.com/absolum/30min" target="_blank">[https://cal.com/absolum/30min](https://cal.com/absolum/30min)</a>
  </Card>
</CardGroup>

Your Claude-based workflows can be running on Emby models **in under 5 minutes**.