---
license: apache-2.0
library_name: transformers
pipeline_tag: image-text-to-text
tags:
- vision
- multimodal
- qwen3.5
- omni
- gguf
- safetensors
base_model:
- Qwen/Qwen3.5-27B
language:
- en
- zh
- multilingual
---

# MARTHA-27B-Qwen3.5-Omni

**Vision + Language AI — 27B parameters. Sees images, reads text, answers everything.**

Built by [Zero Point Intelligence Ltd](https://zeropointai.uk), Dundee, Scotland.

## About

Derivative of [Qwen3.5-27B](https://huggingface.co/Qwen/Qwen3.5-27B) by Alibaba Cloud.
Ghost-passed to produce mathematically unique weights.
See `integrity_manifest.json` for SHA256 proof.

## Vision

Ship two files together for image understanding:
- `MARTHA-27B-Q4_K_M.gguf` (brain)
- `mmproj-f16.gguf` (eyes)

```bash
llama-server -m MARTHA-27B-Q4_K_M.gguf --mmproj mmproj-f16.gguf -ngl 99
```

## License

Apache 2.0 derivative work. Original model by [Alibaba Cloud / Qwen](https://huggingface.co/Qwen).
Copyright 2026 Zero Point Intelligence Ltd.

https://zeropointai.uk