M2 Ultra 128GB
Apple Silicon
128GB VRAM · 800 GB/s bandwidth · 54.4 FP16 TFLOPS · 120W TDP · $5,499 street price
The M2 Ultra 128GB has 128GB of VRAM with 800 GB/s memory bandwidth and 54.4 TFLOPS FP16 compute. At Q4 quantization, it can comfortably run Gemma 3 4B (168 tok/s), Qwen 2.5 7B (88 tok/s), Llama 3.1 8B (83 tok/s). Models larger than ~218B parameters won't fit even at Q4. Electricity cost is approximately $13/month at 120W TDP.
What LLMs can you run?
| Model | Params | Q4 Weight | Fit | Decode |
|---|---|---|---|---|
| Gemma 3 4B | 4.0B | 2 GB | comfortable | 168 tok/s |
| Qwen 2.5 7B | 7.6B | 4 GB | comfortable | 88 tok/s |
| Llama 3.1 8B | 8.0B | 4 GB | comfortable | 83 tok/s |
| Mistral Small 24B | 24.0B | 12 GB | tight | 28 tok/s |
| Gemma 3 27B | 27.4B | 14 GB | tight | 24 tok/s |
| Qwen 2.5 Coder 32B | 32.5B | 16 GB | tight | 20 tok/s |
| Llama 3.3 70B | 70.6B | 35 GB | tight | 9 tok/s |
| Qwen 2.5 72B | 72.7B | 36 GB | tight | 9 tok/s |
| Llama 3.1 405B | 405B | 202 GB | won't fit | |
| DeepSeek R1 671B | 671B | 336 GB | won't fit |
Similar GPUs
| GPU | VRAM | BW | TFLOPS | TDP |
|---|---|---|---|---|
| M1 Ultra 128GB | 128GB | 800 GB/s | 42.5 | 120W |
| M3 Ultra 128GB | 128GB | 800 GB/s | 65.5 | 120W |
| M4 Max 128GB | 128GB | 546 GB/s | 36.9 | 75W |
| M3 Max 128GB | 128GB | 400 GB/s | 32.8 | 75W |
| GB10 | 128GB | 273 GB/s | 29.7 | 140W |
Compare with another GPU
Select another GPU to compare specs and model performance side by side.