RX 7900 XTX 24GBBest value
C31 tok/s decode
~$999 MSRP
Can it run?
Too heavy
Using Q4_K_M in Ollama
Fit status
Too heavy
Decode
12.6 tok/s
TTFT
15407 ms
Safe context
7K
Memory
26.9 GB / 12.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | F | Too heavy | 12.6 tok/s | 22410 ms | 12K |
| Chat | F | Too heavy | 12.6 tok/s | 8404 ms | 4K |
| Coding | F | Too heavy | 12.6 tok/s | 15407 ms | 7K |
| RAG | F | Too heavy | 12.6 tok/s | 28013 ms | 12K |
| Reasoning | F | Too heavy | 12.6 tok/s | 18208 ms | 7K |
How Qwen 2.5 Coder 32B (32B params) fits at each quantization level on RTX 3500 Ada Laptop 12GB (12.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 12.5 GB | Low | F0 |
Q3_K_S | 3 | 15.7 GB | Low | F0 |
NVFP4 | 4 | 17.9 GB | Medium | F0 |
Q4_K_M | 4 | 19.5 GB | Medium | F0 |
Q5_K_M | 5 | 23.0 GB | High | F0 |
Q6_K | 6 | 26.2 GB | High | F0 |
Q8_0 | 8 | 34.2 GB | Very High | F0 |
F16 | 16 | 65.6 GB | Maximum | F0 |
Upgrade options
~$999 MSRP
~$1,099 MSRP
NVIDIA A100 40GBBiggest leap~$10,000 MSRP
RTX PRO 5000 Blackwell 48GBNVIDIA upgrade