M4 Pro 24GB
Apple Silicon
24GB VRAM · 273 GB/s bandwidth · 18.4 FP16 TFLOPS · 45W TDP · $1,599 street price
The M4 Pro 24GB has 24GB of VRAM with 273 GB/s memory bandwidth and 18.4 TFLOPS FP16 compute. At Q4 quantization, it can comfortably run Gemma 3 4B (60 tok/s), Qwen 2.5 7B (31 tok/s). Models larger than ~41B parameters won't fit even at Q4. Electricity cost is approximately $5/month at 45W TDP.
What LLMs can you run?
| Model | Params | Q4 Weight | Fit | Decode |
|---|---|---|---|---|
| Gemma 3 4B | 4.0B | 2 GB | comfortable | 60 tok/s |
| Qwen 2.5 7B | 7.6B | 4 GB | comfortable | 31 tok/s |
| Llama 3.1 8B | 8.0B | 4 GB | tight | 29 tok/s |
| Mistral Small 24B | 24.0B | 12 GB | tight | 10 tok/s |
| Gemma 3 27B | 27.4B | 14 GB | tight | 8 tok/s |
| Qwen 2.5 Coder 32B | 32.5B | 16 GB | tight | 7 tok/s |
| Llama 3.3 70B | 70.6B | 35 GB | won't fit | |
| Qwen 2.5 72B | 72.7B | 36 GB | won't fit | |
| 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 |
|---|---|---|---|---|
| Tesla P40 | 24GB | 347 GB/s | 0.2 | 250W |
| M4 24GB | 24GB | 120 GB/s | 9.2 | 20W |
| RTX 4500 Ada Generation | 24GB | 432 GB/s | 39.6 | 210W |
| Quadro P6000 | 24GB | 432 GB/s | 0.2 | 250W |
| M2 24GB | 24GB | 100 GB/s | 7.2 | 15W |
Compare with another GPU
Select another GPU to compare specs and model performance side by side.