MacBook Pro M3 Pro 18GBBudget pick
C105.6 tok/s decode
~$1,999 MSRP
Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
211.1 tok/s
TTFT
917 ms
Safe context
33K
Memory
4.2 GB / 12.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 211.1 tok/s | 1334 ms | 33K |
| Chat | C | Runs well | 211.1 tok/s | 500 ms | 23K |
| Coding | C | Runs well | 211.1 tok/s | 917 ms | 33K |
| RAG | C | Runs well | 211.1 tok/s | 1668 ms | 33K |
| Reasoning | C | Runs well | 211.1 tok/s | 1084 ms | 33K |
How Qwen 3 1.7B (1.7000000476837158B params) fits at each quantization level on Intel Arc B580 12GB (12.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 0.7 GB | Low | D29 |
Q3_K_S | 3 | 0.8 GB | Low | D29 |
NVFP4 | 4 | 1.0 GB | Medium | D30 |
Q4_K_M | 4 | 1.0 GB | Medium | D30 |
Q5_K_M | 5 | 1.2 GB | High | D30 |
Q6_K | 6 | 1.4 GB | High | D30 |
Q8_0 | 8 | 1.8 GB | Very High | D31 |
F16Best for your GPU | 16 | 3.5 GB | Maximum | D34 |
ollama run qwen-3-1.7bhuggingface-cli download qwen-3-1.7bUpgrade options
~$1,999 MSRP