---
license: apache-2.0
language:
- en
base_model:
- DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED
base_model_relation: quantized
quantized_by: Atomic-Germ
pipeline_tag: text-generation
tags:
- safetensors
- qwen3_5_text
- DuoNeural
- cybersecurity
- agentic
- qlora
- unsloth
- qwen3_5
- obliterated
- terminal
- cli
- reasoning
- function-calling
- tool-use
- text-generation
- conversational
- en
- dataset:oi-uae/cyber-security
- dataset:hotdogs/uka-cyber-dataset
- dataset:trend-cybertron/Primus-Reasoning
- dataset:Lite-Coder/LiteCoder-Terminal-SFT
- dataset:rajistics/openhands-synthetic-conversations
- dataset:emirkaanozdemr/bash_command_data_6K
- dataset:NousResearch/hermes-function-calling-v1
- dataset:open-thoughts/OpenThoughts3-1.2M
- base_model:OBLITERATUS/Ornith-1.5-9B-OBLITERATED
- base_model:finetune:OBLITERATUS/Ornith-1.5-9B-OBLITERATED
- license:apache-2.0
- region:us
---
# Cyber-Ornith-1.5-9B-NPU2

**OpenFlowLM Q4NX conversion of [`DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED`](https://huggingface.co/DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED)** for AMD XDNA NPU inference.

This repository contains a quantized **Q4NX** port of the model, compiled for the OpenFlowLM (OFLM) runtime. It is **not** a GGUF file.

| Item | Value |
|------|-------|
| Source model | [`DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED`](https://huggingface.co/DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED) |
| Weights | `model.q4nx` (7.11 GB) |
| Modality | language |
| OFLM version | `0.1.0` |
| Converted | 2026-10-03 |

## Install and run

This repository works with `oflm-add`, a small installer that copies the model
into the OpenFlowLM user directory and registers the tag. It never
modifies the system OpenFlowLM install.

`pip install oflm-add` or `uv tool install oflm-add`

```bash
uv tool install oflm-add
oflm-add Atomic-Germ/Cyber-Ornith-1.5-9B-NPU2 --family qwen3.5 --xclbin-from Cyber-Ornith-1.5-9B-NPU2
OFLM_CONFIG_PATH="$HOME/.config/oflm/model_list.json" OFLM_XCLBIN_PATH="$HOME/.config/oflm" oflm run Cyber-Ornith-1.5-9B-NPU2
```

## Files

| File | Description |
|------|-------------|
| `model.q4nx` | Quantized weights (Q8_0 / Q4_1 / BF16) |
| `config.json` | OFLM runtime configuration |
| `tokenizer.json` | Tokenizer vocabulary |
| `tokenizer_config.json` | Tokenizer configuration |
| `chat_template.jinja` | Chat template |


---

## Source model card

See the original model card: [DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED](https://huggingface.co/DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED)