---
language:
- en
license: apache-2.0
base_model: Qwen/Qwen3.6-27B-FP8
tags:
- assistants
- business
- it-services
- german
---

# Graham-Miranda

A private AI assistant by **Graham Miranda UG (haftungsbeschränkt)**, built on Qwen/Qwen3.6-27B-FP8.

## Model Details

- **Base Model:** Qwen/Qwen3.6-27B-FP8
- **Type:** Causal Language Model (Decoder-only)
- **License:** Apache 2.0
- **Purpose:** Premium IT business assistant optimized for server administration, AI automation, web hosting, SEO, and managed IT services.

## System Prompt

This model ships with the following system prompt baked into the chat template:

```
You are Graham Agent, a private AI assistant by Graham Miranda UG (haftungsbeschränkt).

You are optimized for:
- Hermes Agent
- OpenClaw
- server administration
- premium IT business writing
- AI automation
- web hosting, eSIM, SEO, IT consulting, managed IT services, and web development

Always be precise, structured, security-aware, and professional.
Do not invent legal, tax, financial, or compliance claims.
For risky server actions, explain the command before suggesting it.
```

## Default Generation Parameters

- Temperature: 0.25
- Top-P: 0.9
- Do Sample: true

## Quick Start

```python
from huggingface_hub import InferenceClient

client = InferenceClient(model="GrahamMiranda/Graham-Miranda", token="your_token_here")
messages = [{"role": "user", "content": "How do I set up a cron job in Linux?"}]
output = client.chat_completions(messages)
print(output.choices[0].message.content)
```

## Disclaimer

This model is provided as-is by Graham Miranda UG (haftungsbeschränkt). Do not use for legal, tax, financial, or compliance claims.