# Can the Instinct MI50 32GB run Hemmingway-1.5?

> Yes. The Instinct MI50 32GB runs Hemmingway-1.5 fully in VRAM at Q6_K (about 26 GB of 32 GB), at roughly 25 tokens/s.

Instinct MI50 32GB: 32 GB (32 GB usable), 1024 GB/s. Estimates at an 8K context.

## Every quantisation

| Quant | Quality | ≈ GB | Fit | Est. tok/s |
|---|---|---|---|---|
| Q2_K | Heavy loss | 13.08 | fits | 64 |
| Q3_K_M | Noticeable loss | 15.6 | fits | 43 |
| Q4_K_M | Small loss | 19.55 | fits | 35 |
| Q5_K_M | Near-lossless | 22.96 | fits | 29 |
| Q6_K (pick) | Near-lossless | 25.83 | fits | 25 |
| Q8_0 | Lossless | 30.98 | tight | 20 |
| F16 | Lossless | 55.95 | offload | 1.3 |

## Run it

Q6_K straight from Hugging Face (ollama):

```bash
ollama run hf.co/bartowski/HemmingwayAI_Hemmingway-1.5-GGUF:Q6_K
```

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

```bash
llama-server -hf bartowski/HemmingwayAI_Hemmingway-1.5-GGUF:Q6_K
```

Docker with Vulkan (llama.cpp):

```bash
docker run -d --name llama --device /dev/dri -p 8080:8080 \
  -v ~/.cache/llama.cpp:/root/.cache/llama.cpp \
  ghcr.io/ggml-org/llama.cpp:server-vulkan \
  -hf bartowski/HemmingwayAI_Hemmingway-1.5-GGUF:Q6_K --host 0.0.0.0 --port 8080 -ngl 99
```

Download with the lms CLI (lmstudio):

```bash
lms get bartowski/HemmingwayAI_Hemmingway-1.5-GGUF@Q6_K
```

Chat with the pipeline API (transformers):

```python
from transformers import pipeline

pipe = pipeline("text-generation", model="HemmingwayAI/Hemmingway-1.5", 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

- [openai/gpt-oss-20b](https://selfhost.directory/models/openai/gpt-oss-20b/on/instinct-mi50): Q8_0 ≈12.72 GB, est. 126 tok/s
- [zai-org/GLM-4.7-Flash](https://selfhost.directory/models/zai-org/glm-4.7-flash/on/instinct-mi50): Q5_K_M ≈28.78 GB, est. 172 tok/s
- [Qwen/Qwen3-32B](https://selfhost.directory/models/qwen/qwen3-32b/on/instinct-mi50): Q6_K ≈28.79 GB, est. 21 tok/s
- [ornith-ai/Ornith-1.5-35B-A3B](https://selfhost.directory/models/ornith-ai/ornith-1.5-35b-a3b/on/instinct-mi50): Q5_K_M ≈26.06 GB, est. 264 tok/s
- [Qwen/Qwen3-30B-A3B](https://selfhost.directory/models/qwen/qwen3-30b-a3b/on/instinct-mi50): Q6_K ≈25.79 GB, est. 228 tok/s
- [LiquidAI/LFM2.5-24B-A2B](https://selfhost.directory/models/liquidai/lfm2.5-24b-a2b/on/instinct-mi50): Q8_0 ≈26.06 GB, est. 265 tok/s

## Other GPUs that run it

- [Tesla T4](https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/tesla-t4): Q3_K_M, est. 13 tok/s
- [Arc A770 16GB](https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/arc-a770): Q3_K_M, est. 23 tok/s
- [Radeon RX 7800 XT](https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/rx-7800-xt): Q3_K_M, est. 26 tok/s
- [Tesla P40](https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/tesla-p40): Q4_K_M, est. 12 tok/s
- [Radeon RX 7900 XTX](https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/rx-7900-xtx): Q4_K_M, est. 32 tok/s
- [GeForce RTX 5090](https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/rtx-5090): Q6_K, est. 44 tok/s
- [Apple M4 Pro (48 GB)](https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/apple-m4-pro-48gb): Q8_0, est. 5.3 tok/s
- [RTX A6000](https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/rtx-a6000): Q8_0, est. 15 tok/s

---
Model: https://selfhost.directory/models/hemmingwayai/hemmingway-1.5. GPU: https://selfhost.directory/models/gpus/instinct-mi50. Source: https://selfhost.directory/models/hemmingwayai/hemmingway-1.5/on/instinct-mi50