MacBook Pro M3 Max 128GBBest value
D3 tok/s decode
~$2,499 MSRP
Can it run?
Too heavy
Using Q4_K_M in vLLM
Fit status
Too heavy
Decode
4.5 tok/s
TTFT
42563 ms
Safe context
4K
Memory
99.0 GB / 24.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | F | Too heavy | 4.5 tok/s | 61909 ms | 6K |
| Chat | F | Too heavy | 4.5 tok/s | 23216 ms | 4K |
| Coding | F | Too heavy | 4.5 tok/s | 42563 ms | 4K |
| RAG | F | Too heavy | 4.5 tok/s | 77387 ms | 6K |
| Reasoning | F | Too heavy | 4.5 tok/s | 50301 ms | 4K |
How Devstral 2 123B Instruct (123B params) fits at each quantization level on RTX 4500 Ada 24GB (24.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 48.0 GB | Low | F0 |
Q3_K_S | 3 | 60.3 GB | Low | F0 |
NVFP4 | 4 |
Upgrade options
~$2,499 MSRP
~$8,000 MSRP
NVIDIA B200 180GBBiggest leap68.9 GB |
| Medium |
| F0 |
Q4_K_M | 4 | 75.0 GB | Medium | F0 |
Q5_K_M | 5 | 88.6 GB | High | F0 |
Q6_K | 6 | 100.9 GB | High | F0 |
Q8_0 | 8 | 131.6 GB | Very High | F0 |
F16 | 16 | 252.2 GB | Maximum | F0 |
~$30,000 MSRP
NVIDIA H200 141GBNVIDIA upgrade~$30,000 MSRP