RX 7900 XTX 24GBBudget pick
C51.5 tok/s decode
~$999 MSRP
Can it run?
Tight fit
Using Q4_K_M in Ollama
Fit status
Tight fit
Decode
6.4 tok/s
TTFT
30071 ms
Safe context
17K
Memory
21.5 GB / 23.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | D | Very compromised (needs ~1 GB host RAM) | 6.3 tok/s | 44768 ms | 30K |
| Chat | C | Tight fit | 6.4 tok/s | 16402 ms | 9K |
| Coding | C | Tight fit | 6.4 tok/s | 30071 ms | 17K |
| RAG | D | Very compromised (needs ~1 GB host RAM) | 6.3 tok/s | 55960 ms | 30K |
| Reasoning | C | Tight fit | 6.4 tok/s | 35538 ms | 17K |
How Codestral 22B v0.1 (22B params) fits at each quantization level on MacBook Pro M4 32GB (23.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 8.6 GB | Low | D37 |
Q3_K_S | 3 | 10.8 GB | Low | D39 |
NVFP4 | 4 | 12.3 GB | Medium | C41 |
Q4_K_M | 4 | 13.4 GB | Medium | C42 |
Q5_K_MBest for your GPU | 5 | 15.8 GB | High | C44 |
Q6_K | 6 | 18.0 GB | High | C44 |
Q8_0 | 8 | 23.5 GB | Very High | F0 |
F16 | 16 | 45.1 GB | Maximum | F0 |
Upgrade options
~$999 MSRP
RTX 3090 24GBBest value~$1,499 MSRP
RTX 5090 32GBBiggest leap~$1,999 MSRP
~$2,499 MSRP