# Can the Apple M4 (16 GB) run Bielik-11B-v3.1-Instruct?

> Yes. The Apple M4 (16 GB) runs Bielik-11B-v3.1-Instruct fully in unified memory at Q5_K_M (about 9.4 GB of 11 GB), at roughly 8.3 tokens/s.

Apple M4 (16 GB): 16 GB (10.7 GB usable), 120 GB/s. Estimates at an 8K context.

## Every quantisation

| Quant | Quality | ≈ GB | Fit | Est. tok/s |
|---|---|---|---|---|
| Q2_K | Heavy loss | 5.3 | fits | 18 |
| Q3_K_M | Noticeable loss | 7.05 | fits | 12 |
| Q4_K_M | Small loss | 8.28 | fits | 9.7 |
| Q5_K_M (pick) | Near-lossless | 9.44 | fits | 8.3 |
| Q6_K | Near-lossless | 10.67 | tight | 7.2 |
| Q8_0 | Lossless | 13.31 | no | — |
| F16 | Lossless | 23.56 | no | — |

## Run it

Apple Silicon with mlx-lm (mlx):

```bash
pip install mlx-lm
mlx_lm.generate --model speakleash/Bielik-11B-v3.1-Instruct-MLX-4bit --prompt "Hello"
```

OpenAI-compatible server (mlx):

```bash
mlx_lm.server --model speakleash/Bielik-11B-v3.1-Instruct-MLX-4bit --port 8080
```

Q5_K_M straight from Hugging Face (ollama):

```bash
ollama run hf.co/speakleash/Bielik-11B-v3.1-Instruct-GGUF:Q5_K_M
```

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

```bash
llama-server -hf speakleash/Bielik-11B-v3.1-Instruct-GGUF:Q5_K_M
```

Download with the lms CLI (lmstudio):

```bash
lms get speakleash/Bielik-11B-v3.1-Instruct-GGUF@Q5_K_M
```

Chat with the pipeline API (transformers):

```python
from transformers import pipeline

pipe = pipeline("text-generation", model="speakleash/Bielik-11B-v3.1-Instruct", 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

- [xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic](https://selfhost.directory/models/xtools-at/deepseek-v4-pro-qwen3.5-9b-uncensored-heretic/on/apple-m4-16gb): Q6_K ≈8.89 GB, est. 8.5 tok/s
- [Qwen/Qwen3-8B](https://selfhost.directory/models/qwen/qwen3-8b/on/apple-m4-16gb): Q8_0 ≈10.15 GB, est. 7.6 tok/s
- [meta-llama/Llama-3.1-8B-Instruct](https://selfhost.directory/models/meta-llama/llama-3.1-8b-instruct/on/apple-m4-16gb): Q8_0 ≈9.84 GB, est. 7.7 tok/s
- [google/gemma-2-9b-it](https://selfhost.directory/models/google/gemma-2-9b-it/on/apple-m4-16gb): Q5_K_M ≈9.63 GB, est. 10 tok/s
- [ornith-ai/Ornith-1.5-9B](https://selfhost.directory/models/ornith-ai/ornith-1.5-9b/on/apple-m4-16gb): Q6_K ≈9.03 GB, est. 8.3 tok/s
- [meta-llama/Meta-Llama-3-8B-Instruct](https://selfhost.directory/models/meta-llama/meta-llama-3-8b-instruct/on/apple-m4-16gb): Q8_0 ≈9.85 GB, est. 7.7 tok/s

## Other GPUs that run it

- [GeForce RTX 4060](https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/rtx-4060): Q3_K_M, est. 27 tok/s
- [Radeon RX 6700 XT](https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/rx-6700-xt): Q6_K, est. 23 tok/s
- [GeForce RTX 3060 12GB](https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/rtx-3060): Q6_K, est. 22 tok/s
- [Arc B580](https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/arc-b580): Q6_K, est. 27 tok/s
- [Tesla T4](https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/tesla-t4): Q8_0, est. 15 tok/s
- [Arc A770 16GB](https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/arc-a770): Q8_0, est. 26 tok/s
- [Radeon RX 7800 XT](https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/rx-7800-xt): Q8_0, est. 29 tok/s
- [Tesla P40](https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/tesla-p40): Q8_0, est. 16 tok/s

---
Model: https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct. GPU: https://selfhost.directory/models/gpus/apple-m4-16gb. Source: https://selfhost.directory/models/speakleash/bielik-11b-v3.1-instruct/on/apple-m4-16gb