← Back to comparison
APPLE
Mac Studio M4 Ultra 192GB
M4 Ultra · M4 · ACTIVE
We earn commissions from purchases made through links on this site. This doesn't affect our rankings or recommendations.
Prices updated 5 days ago
Specifications
| VRAM | 192 GB |
| Memory type | Unified |
| Memory bandwidth | 819 GB/s |
| CPU cores | 32 |
| GPU cores | 80 |
| TDP | 200W |
| Form factor | Mac Studio |
| Max model (Q4) | ~380B parameters |
Inference Benchmarks (Q4_K_M)
Llama 3.3 8B
85.0 tok/s
Qwen 3 32B
38.0 tok/s
Llama 3.3 70B
18.0 tok/s
llama.cpp, batch_size=1, ctx=4096, single GPU.
What Can You Run?
| Model | Q4_K_M | Q8_0 | FP16 |
|---|---|---|---|
| Llama 3.3 8B8B | Excellent~85 tok/s | Usable | Usable |
| Llama 3.3 70B70.6B | Usable~18 tok/s | Usable | Usable |
| Qwen 3 8B8.2B | Usable | Usable | Usable |
| Qwen 3 32B32.8B | Good~38 tok/s | Usable | Usable |
| DeepSeek R1 70B70.6B | Usable | Usable | Usable |
| Mistral Nemo 12B12.2B | Usable | Usable | Usable |
| Phi-4 14B14B | Usable | Usable | Usable |
| Gemma 3 27B27.4B | Usable | Usable | Usable |
| Codestral 25B25.3B | Usable | Usable | Usable |
| Command R 35B35B | Usable | Usable | Usable |
Notes
Can run virtually any open model at good quantization. The ultimate local LLM machine.