# Can the NVIDIA H100 SXM 80GB run plamo-3-translate-31b?

> Yes. The NVIDIA H100 SXM 80GB runs plamo-3-translate-31b fully in VRAM at FP16 / BF16 (about 71 GB of 80 GB).

NVIDIA H100 SXM 80GB: 80 GB (80 GB usable), 3350 GB/s. Estimates at an 8K context.

## Every quantisation

| Quant | Quality | ≈ GB | Fit | Est. tok/s |
|---|---|---|---|---|
| Q4_K_M | Small loss | 22.33 | fits | — |
| Q8_0 | Lossless | 38.21 | fits | — |
| F16 (pick) | Lossless | 71.48 | fits | — |

## Run it

Q6_K straight from Hugging Face (ollama):

```bash
ollama run hf.co/mradermacher/plamo-3-translate-31b-i1-GGUF:Q6_K
```

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

```bash
llama-server -hf mradermacher/plamo-3-translate-31b-i1-GGUF:Q6_K
```

Download with the lms CLI (lmstudio):

```bash
lms get mradermacher/plamo-3-translate-31b-i1-GGUF@Q6_K
```

Serve an OpenAI-compatible API (vllm):

```bash
vllm serve pfnet/plamo-3-translate-31b
```

Smaller FP8 build (vllm):

```bash
vllm serve pfnet/plamo-3-translate-31b-FP8
```

Docker Compose (NVIDIA) (vllm):

```yaml
services:
  vllm:
    image: vllm/vllm-openai:latest
    command: pfnet/plamo-3-translate-31b
    ports:
      - "8000:8000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
```

Run with the translation pipeline (transformers):

```python
from transformers import pipeline

pipe = pipeline("translation", model="pfnet/plamo-3-translate-31b")
result = pipe("Self-hosting puts you in control of your data.")
print(result)
```


## Translation models that fit comfortably

- [tencent/Hy-MT2-30B-A3B](https://selfhost.directory/models/tencent/hy-mt2-30b-a3b/on/h100-sxm): F16 ≈67.7 GB
- [tencent/Hy-SimulMT-30B-A3B](https://selfhost.directory/models/tencent/hy-simulmt-30b-a3b/on/h100-sxm): F16 ≈68.44 GB
- [facebook/nllb-moe-54b](https://selfhost.directory/models/facebook/nllb-moe-54b/on/h100-sxm): Q8_0 ≈64.62 GB
- [IndexTeam/Index-Translate-9B](https://selfhost.directory/models/indexteam/index-translate-9b/on/h100-sxm): F16 ≈22.08 GB
- [tencent/Hy-MT2-7B](https://selfhost.directory/models/tencent/hy-mt2-7b/on/h100-sxm): F16 ≈18.45 GB
- [tencent/Hunyuan-MT-7B](https://selfhost.directory/models/tencent/hunyuan-mt-7b/on/h100-sxm): F16 ≈18.45 GB

## Other GPUs that run it

- [Tesla P40](https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/tesla-p40): Q4_K_M
- [Radeon RX 7900 XTX](https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/rx-7900-xtx): Q4_K_M
- [Instinct MI50 32GB](https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/instinct-mi50): Q4_K_M
- [GeForce RTX 5090](https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/rtx-5090): Q4_K_M
- [Apple M4 Pro (48 GB)](https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/apple-m4-pro-48gb): Q4_K_M
- [RTX A6000](https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/rtx-a6000): Q8_0
- [Apple M1 Max (64 GB)](https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/apple-m1-max-64gb): Q8_0
- [NVIDIA A100 80GB](https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/a100-80gb): F16

---
Model: https://selfhost.directory/models/pfnet/plamo-3-translate-31b. GPU: https://selfhost.directory/models/gpus/h100-sxm. Source: https://selfhost.directory/models/pfnet/plamo-3-translate-31b/on/h100-sxm