# Can the NVIDIA A100 80GB run MLExpert?

> Yes. The NVIDIA A100 80GB runs MLExpert fully in VRAM at FP16 / BF16 (about 69 GB of 80 GB), at roughly 162 tokens/s.

NVIDIA A100 80GB: 80 GB (80 GB usable), 2039 GB/s. Estimates at an 8K context.

## Every quantisation

| Quant | Quality | ≈ GB | Fit | Est. tok/s |
|---|---|---|---|---|
| Q2_K | Heavy loss | 14.66 | fits | 987 |
| Q3_K_M | Noticeable loss | 18.09 | fits | 664 |
| Q4_K_M | Small loss | 22.96 | fits | 535 |
| Q5_K_M | Near-lossless | 27.52 | fits | 456 |
| Q6_K | Near-lossless | 31.19 | fits | 394 |
| Q8_0 | Lossless | 37.23 | fits | 305 |
| F16 (pick) | Lossless | 68.97 | fits | 162 |

## Run it

BF16 straight from Hugging Face (ollama):

```bash
ollama run hf.co/bartowski/InternScience_MLExpert-GGUF:BF16
```

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

```bash
llama-server -hf bartowski/InternScience_MLExpert-GGUF:BF16
```

Download with the lms CLI (lmstudio):

```bash
lms get bartowski/InternScience_MLExpert-GGUF@BF16
```

Serve an OpenAI-compatible API (vllm):

```bash
vllm serve InternScience/MLExpert
```

Docker Compose (NVIDIA) (vllm):

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

Serve an OpenAI-compatible API (sglang):

```bash
sglang serve InternScience/MLExpert --host 0.0.0.0 --port 30000
```

Docker Compose (NVIDIA) (sglang):

```yaml
services:
  sglang:
    image: lmsysorg/sglang:latest
    command: sglang serve InternScience/MLExpert --host 0.0.0.0 --port 30000
    ports:
      - "30000:30000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
```

Chat with the pipeline API (transformers):

```python
from transformers import pipeline

pipe = pipeline("text-generation", model="InternScience/MLExpert", dtype="auto", device_map="auto")
messages = [{"role": "user", "content": "Explain self-hosting in one paragraph."}]
out = pipe(messages, max_new_tokens=512)
print(out[0]["generated_text"][-1]["content"])
```


## Chat models that fit comfortably

- [ornith-ai/Ornith-1.5-35B-A3B](https://selfhost.directory/models/ornith-ai/ornith-1.5-35b-a3b/on/a100-80gb): Q8_0 ≈38.11 GB, est. 352 tok/s
- [zai-org/GLM-4.7-Flash](https://selfhost.directory/models/zai-org/glm-4.7-flash/on/a100-80gb): Q8_0 ≈38.98 GB, est. 229 tok/s
- [Qwen/Qwen3-32B](https://selfhost.directory/models/qwen/qwen3-32b/on/a100-80gb): Q8_0 ≈36.55 GB, est. 32 tok/s
- [RecursiveSI/flint](https://selfhost.directory/models/recursivesi/flint/on/a100-80gb): Q8_0 ≈42.64 GB, est. 844 tok/s
- [openai/gpt-oss-20b](https://selfhost.directory/models/openai/gpt-oss-20b/on/a100-80gb): Q8_0 ≈12.72 GB, est. 252 tok/s
- [Qwen/Qwen3-30B-A3B](https://selfhost.directory/models/qwen/qwen3-30b-a3b/on/a100-80gb): Q8_0 ≈33.01 GB, est. 352 tok/s

## Other GPUs that run it

- [Tesla P40](https://selfhost.directory/models/internscience/mlexpert/on/tesla-p40): Q3_K_M, est. 113 tok/s
- [Radeon RX 7900 XTX](https://selfhost.directory/models/internscience/mlexpert/on/rx-7900-xtx): Q3_K_M, est. 313 tok/s
- [Instinct MI50 32GB](https://selfhost.directory/models/internscience/mlexpert/on/instinct-mi50): Q5_K_M, est. 229 tok/s
- [GeForce RTX 5090](https://selfhost.directory/models/internscience/mlexpert/on/rtx-5090): Q5_K_M, est. 401 tok/s
- [Apple M4 Pro (48 GB)](https://selfhost.directory/models/internscience/mlexpert/on/apple-m4-pro-48gb): Q6_K, est. 53 tok/s
- [RTX A6000](https://selfhost.directory/models/internscience/mlexpert/on/rtx-a6000): Q8_0, est. 115 tok/s
- [Apple M1 Max (64 GB)](https://selfhost.directory/models/internscience/mlexpert/on/apple-m1-max-64gb): Q8_0, est. 60 tok/s
- [NVIDIA H100 SXM 80GB](https://selfhost.directory/models/internscience/mlexpert/on/h100-sxm): F16, est. 267 tok/s

---
Model: https://selfhost.directory/models/internscience/mlexpert. GPU: https://selfhost.directory/models/gpus/a100-80gb. Source: https://selfhost.directory/models/internscience/mlexpert/on/a100-80gb