# Can I run DeepSeek R1 on Apple M3 Ultra (256GB)?

> **No, not enough memory.** Needs ~383.7 GB even at Q4_K_M, but only ~192 GB is usable.

Last validated: 2026-06-15.

## Memory math
- DeepSeek R1: 671B (MoE, 37B active); Q4_K_M weighs 376.66 GB on disk.
- At Q4_K_M with a 4k context it needs ~383.7 GB (weights + KV cache + ~0.8 GB runtime overhead, ±15% with context length).
- Apple M3 Ultra (256GB): 256 GB unified, ~192 GB usable for model weights.
- Headroom: ~-191.7 GB.

## How to run
Recommended quant: n/a. Best tool on macOS: LM Studio.

```sh
ollama run deepseek-r1:671b
```

## Other models that run on Apple M3 Ultra (256GB)
- [Qwen3 235B A22B (235B)](https://localmodel.run/can-i-run/qwen3-235b-a22b/apple-m3-ultra-256gb)
- [gpt-oss 120B (117B)](https://localmodel.run/can-i-run/gpt-oss-120b/apple-m3-ultra-256gb)
- [Llama 4 Scout (109B)](https://localmodel.run/can-i-run/llama-4-scout/apple-m3-ultra-256gb)
- [Sarvam-105B (105B)](https://localmodel.run/can-i-run/sarvam-105b/apple-m3-ultra-256gb)
- [Qwen2.5 72B (72B)](https://localmodel.run/can-i-run/qwen2.5-72b/apple-m3-ultra-256gb)
- [Llama 3.3 70B (70B)](https://localmodel.run/can-i-run/llama-3.3-70b/apple-m3-ultra-256gb)

Estimate, not a guarantee. Sources: https://ollama.com/library/deepseek-r1, https://ollama.com/library/deepseek-r1/tags, https://huggingface.co/unsloth/DeepSeek-R1-GGUF, https://huggingface.co/deepseek-ai/DeepSeek-R1.
More: https://localmodel.run/can-i-run/deepseek-r1/apple-m3-ultra-256gb