# Can the Radeon RX 6700 XT run plamo-3-translate-8b?

> Yes. The Radeon RX 6700 XT runs plamo-3-translate-8b fully in VRAM at Q8_0 (about 11 GB of 12 GB).

Radeon RX 6700 XT: 12 GB (12 GB usable), 384 GB/s. Estimates at an 8K context.

## Every quantisation

| Quant | Quality | ≈ GB | Fit | Est. tok/s |
|---|---|---|---|---|
| Q4_K_M | Small loss | 6.4 | fits | — |
| Q8_0 (pick) | Lossless | 10.63 | fits | — |
| F16 | Lossless | 19.57 | no | — |

## Run it

Q6_K straight from Hugging Face (ollama):

```bash
ollama run hf.co/mradermacher/plamo-3-translate-8b-i1-GGUF:Q6_K
```

Docker with ROCm (AMD) (ollama):

```bash
docker run -d --name ollama --device /dev/kfd --device /dev/dri \
  -e HSA_OVERRIDE_GFX_VERSION=10.3.0 \
  -v ollama:/root/.ollama -p 11434:11434 ollama/ollama:rocm
docker exec -it ollama ollama run hf.co/mradermacher/plamo-3-translate-8b-i1-GGUF:Q6_K
```

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

```bash
llama-server -hf mradermacher/plamo-3-translate-8b-i1-GGUF:Q6_K
```

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

```bash
docker run -d --name llama --device /dev/kfd --device /dev/dri -p 8080:8080 \
  -e HSA_OVERRIDE_GFX_VERSION=10.3.0 \
  -v ~/.cache/llama.cpp:/root/.cache/llama.cpp \
  ghcr.io/ggml-org/llama.cpp:server-rocm \
  -hf mradermacher/plamo-3-translate-8b-i1-GGUF:Q6_K --host 0.0.0.0 --port 8080 -ngl 99
```

Download with the lms CLI (lmstudio):

```bash
lms get mradermacher/plamo-3-translate-8b-i1-GGUF@Q6_K
```

Serve an OpenAI-compatible API (vllm):

```bash
vllm serve pfnet/plamo-3-translate-8b
```

Smaller FP8 build (vllm):

```bash
vllm serve pfnet/plamo-3-translate-8b-FP8
```

Docker Compose (AMD ROCm) (vllm):

```yaml
services:
  vllm:
    image: rocm/vllm:latest
    command: vllm serve pfnet/plamo-3-translate-8b --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
```

Run with the translation pipeline (transformers):

```python
from transformers import pipeline

pipe = pipeline("translation", model="pfnet/plamo-3-translate-8b")
result = pipe("Self-hosting puts you in control of your data.")
print(result)
```


## Translation models that fit comfortably

- [tencent/Hy-MT2-7B](https://selfhost.directory/models/tencent/hy-mt2-7b/on/rx-6700-xt): Q8_0 ≈10.03 GB
- [tencent/Hunyuan-MT-7B](https://selfhost.directory/models/tencent/hunyuan-mt-7b/on/rx-6700-xt): Q8_0 ≈10.04 GB
- [IndexTeam/Index-Translate-9B](https://selfhost.directory/models/indexteam/index-translate-9b/on/rx-6700-xt): Q4_K_M ≈7.04 GB
- [ByteDance-Seed/Seed-X-PPO-7B](https://selfhost.directory/models/bytedance-seed/seed-x-ppo-7b/on/rx-6700-xt): Q8_0 ≈9.42 GB
- [tencent/HY-MT1.5-7B](https://selfhost.directory/models/tencent/hy-mt1.5-7b/on/rx-6700-xt): Q8_0 ≈10.03 GB
- [ByteDance-Seed/Seed-X-Instruct-7B](https://selfhost.directory/models/bytedance-seed/seed-x-instruct-7b/on/rx-6700-xt): Q8_0 ≈9.42 GB

## Other GPUs that run it

- [GeForce RTX 4060](https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/rtx-4060): Q4_K_M
- [Apple M4 (16 GB)](https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/apple-m4-16gb): Q4_K_M
- [GeForce RTX 3060 12GB](https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/rtx-3060): Q8_0
- [Arc B580](https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/arc-b580): Q8_0
- [Tesla T4](https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/tesla-t4): Q8_0
- [Arc A770 16GB](https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/arc-a770): Q8_0
- [Radeon RX 7800 XT](https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/rx-7800-xt): Q8_0
- [Tesla P40](https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/tesla-p40): F16

---
Model: https://selfhost.directory/models/pfnet/plamo-3-translate-8b. GPU: https://selfhost.directory/models/gpus/rx-6700-xt. Source: https://selfhost.directory/models/pfnet/plamo-3-translate-8b/on/rx-6700-xt