# Can the Radeon RX 9070 XT run DeskForge-Gemma4-E4B?

> Yes. The Radeon RX 9070 XT runs DeskForge-Gemma4-E4B fully in VRAM at Q8_0 (about 9.0 GB of 16 GB), at roughly 42 tokens/s.

Radeon RX 9070 XT: 16 GB (16 GB usable), 640 GB/s. Estimates at an 8K context.

## Every quantisation

| Quant | Quality | ≈ GB | Fit | Est. tok/s |
|---|---|---|---|---|
| Q2_K | Heavy loss | 5.44 | fits | 135 |
| Q3_K_M | Noticeable loss | 5.87 | fits | 91 |
| Q4_K_M | Small loss | 6.35 | fits | 73 |
| Q5_K_M | Near-lossless | 6.76 | fits | 62 |
| Q6_K | Near-lossless | 7.2 | fits | 54 |
| Q8_0 (pick) | Lossless | 8.96 | fits | 42 |
| F16 | Lossless | 15.78 | tight | 22 |

## Run it

Q8_0 straight from Hugging Face (ollama):

```bash
ollama run hf.co/mradermacher/DeskForge-Gemma4-E4B-GGUF:Q8_0
```

Docker with ROCm (AMD) (ollama):

```bash
docker run -d --name ollama --device /dev/kfd --device /dev/dri \
  -v ollama:/root/.ollama -p 11434:11434 ollama/ollama:rocm
docker exec -it ollama ollama run hf.co/mradermacher/DeskForge-Gemma4-E4B-GGUF:Q8_0
```

OpenAI-compatible server with a web UI (llama.cpp):

```bash
llama-server -hf mradermacher/DeskForge-Gemma4-E4B-GGUF:Q8_0
```

Docker with ROCm (AMD) (llama.cpp):

```bash
docker run -d --name llama --device /dev/kfd --device /dev/dri -p 8080:8080 \
  -v ~/.cache/llama.cpp:/root/.cache/llama.cpp \
  ghcr.io/ggml-org/llama.cpp:server-rocm \
  -hf mradermacher/DeskForge-Gemma4-E4B-GGUF:Q8_0 --host 0.0.0.0 --port 8080 -ngl 99
```

Download with the lms CLI (lmstudio):

```bash
lms get mradermacher/DeskForge-Gemma4-E4B-GGUF@Q8_0
```

Serve an OpenAI-compatible API (vllm):

```bash
vllm serve docling-project/DeskForge-Gemma4-E4B
```

Docker Compose (AMD ROCm) (vllm):

```yaml
services:
  vllm:
    image: rocm/vllm:latest
    command: vllm serve docling-project/DeskForge-Gemma4-E4B --host 0.0.0.0 --port 8000
    ports:
      - "8000:8000"
    ipc: host
    devices:
      - /dev/kfd
      - /dev/dri
    group_add:
      - video
    security_opt:
      - seccomp=unconfined
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
```

Serve an OpenAI-compatible API (sglang):

```bash
sglang serve docling-project/DeskForge-Gemma4-E4B --host 0.0.0.0 --port 30000
```

Ask about an image (transformers):

```python
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="docling-project/DeskForge-Gemma4-E4B", dtype="auto", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
            {"type": "text", "text": "Describe this image."},
        ],
    }
]
out = pipe(text=messages, max_new_tokens=256)
print(out[0]["generated_text"][-1]["content"])
```


## Vision models that fit comfortably

- [Qwen/Qwen2.5-VL-7B-Instruct](https://selfhost.directory/models/qwen/qwen2.5-vl-7b-instruct/on/rx-9070-xt): Q8_0 ≈8.86 GB, est. 40 tok/s
- [Qwen/Qwen3-VL-8B-Instruct](https://selfhost.directory/models/qwen/qwen3-vl-8b-instruct/on/rx-9070-xt): Q8_0 ≈10.15 GB, est. 38 tok/s
- [Qwen/Qwen3.5-9B](https://selfhost.directory/models/qwen/qwen3.5-9b/on/rx-9070-xt): Q8_0 ≈10.81 GB, est. 34 tok/s
- [XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B](https://selfhost.directory/models/xiaomimimo/mimo-v2.6-distill-qwen-9b/on/rx-9070-xt): Q8_0 ≈10.83 GB, est. 35 tok/s
- [openbmb/MiniCPM-V-4_5](https://selfhost.directory/models/openbmb/minicpm-v-4_5/on/rx-9070-xt): Q8_0 ≈10.15 GB, est. 38 tok/s
- [ByteDance-Seed/UI-TARS-1.5-7B](https://selfhost.directory/models/bytedance-seed/ui-tars-1.5-7b/on/rx-9070-xt): Q8_0 ≈8.86 GB, est. 40 tok/s

## Other GPUs that run it

- [GeForce RTX 4060](https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/rtx-4060): Q6_K, est. 23 tok/s
- [Apple M4 (16 GB)](https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/apple-m4-16gb): Q8_0, est. 7.8 tok/s
- [Radeon RX 6700 XT](https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/rx-6700-xt): Q8_0, est. 25 tok/s
- [GeForce RTX 3060 12GB](https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/rtx-3060): Q8_0, est. 23 tok/s
- [Arc B580](https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/arc-b580): Q8_0, est. 30 tok/s
- [Tesla T4](https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/tesla-t4): Q8_0, est. 21 tok/s
- [Arc A770 16GB](https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/arc-a770): Q8_0, est. 37 tok/s
- [Radeon RX 7800 XT](https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/rx-7800-xt): Q8_0, est. 41 tok/s

---
Model: https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b. GPU: https://selfhost.directory/models/gpus/rx-9070-xt. Source: https://selfhost.directory/models/docling-project/deskforge-gemma4-e4b/on/rx-9070-xt