MacBook Pro M4 32GBBest value
C5.9 tok/s decode
~$799 MSRP
Can it run?
Too heavy
Using Q4_K_M in Ollama
Fit status
Too heavy
Decode
6.3 tok/s
TTFT
30754 ms
Safe context
4K
Memory
20.0 GB / 4.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | F | Too heavy | 4.4 tok/s | 64416 ms | 5K |
| Chat | F | Too heavy | 4.4 tok/s | 24156 ms | 4K |
| Coding | F | Too heavy | 6.3 tok/s | 30754 ms | 4K |
| RAG | F | Too heavy | 4.4 tok/s | 80520 ms | 5K |
| Reasoning | F | Too heavy | 4.4 tok/s | 52338 ms | 4K |
How Devstral Small 2 24B Instruct (24B params) fits at each quantization level on GTX 1650 4GB (4.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 9.4 GB | Low | F0 |
Q3_K_S | 3 | 11.8 GB | Low | F0 |
NVFP4 | 4 |
Upgrade options
~$799 MSRP
~$999 MSRP
RTX 5090 32GBBiggest leap| Medium |
| F0 |
Q4_K_M | 4 | 14.6 GB | Medium | F0 |
Q5_K_M | 5 | 17.3 GB | High | F0 |
Q6_K | 6 | 19.7 GB | High | F0 |
Q8_0 | 8 | 25.7 GB | Very High | F0 |
F16 | 16 | 49.2 GB | Maximum | F0 |
~$1,999 MSRP
RTX PRO 4500 Blackwell 32GBNVIDIA upgrade