MacBook Pro M1 Pro 16GBBudget pick
C106.6 tok/s decode
~$1,999 MSRP
Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
448.5 tok/s
TTFT
432 ms
Safe context
38K
Memory
4.2 GB / 10.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 473.4 tok/s | 595 ms | 76K |
| Chat | C | Runs well | 473.4 tok/s | 350 ms | 19K |
| Coding | C | Runs well | 448.5 tok/s | 432 ms | 38K |
| RAG | C | Runs well | 473.4 tok/s | 743 ms | 76K |
| Reasoning | C | Runs well | 473.4 tok/s | 483 ms | 38K |
How Gemmasutra Mini 2B v1 (2B params) fits at each quantization level on RTX 3080 10GB (10.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 0.8 GB | Low | D31 |
Q3_K_S | 3 | 1.0 GB | Low | D32 |
NVFP4 | 4 |
Upgrade options
~$1,999 MSRP
~$1,999 MSRP
| Medium |
| D32 |
Q4_K_M | 4 | 1.2 GB | Medium | D32 |
Q5_K_M | 5 | 1.4 GB | High | D32 |
Q6_K | 6 | 1.6 GB | High | D33 |
Q8_0 | 8 | 2.1 GB | Very High | D34 |
F16Best for your GPU | 16 | 4.1 GB | Maximum | D38 |