RX 7900 XTX 24GBBudget pick
B65.3 tok/s decode
~$999 MSRP
Can it run?
Runs well
Using Q5_K_M in Ollama
Fit status
Runs well
Decode
8.2 tok/s
TTFT
23725 ms
Safe context
16K
Memory
17.8 GB / 23.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Tight fit | 8.2 tok/s | 34510 ms | 16K |
| Chat | C | Runs well | 8.2 tok/s | 12941 ms | 11K |
| Coding | C | Runs well | 8.2 tok/s | 23725 ms | 16K |
| RAG | C | Tight fit | 8.2 tok/s | 43137 ms | 16K |
| Reasoning | C | Runs well | 8.2 tok/s | 28039 ms | 16K |
How StarCoder2 15B (15B params) fits at each quantization level on MacBook Pro M4 32GB (23.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 5.9 GB | Low | D34 |
Q3_K_S | 3 | 7.4 GB | Low | D36 |
NVFP4 | 4 | 8.4 GB | Medium | D37 |
Q4_K_M | 4 | 9.2 GB | Medium | D37 |
Q5_K_M | 5 | 10.8 GB | High | D39 |
Q6_K | 6 | 12.3 GB | High | C40 |
Q8_0Best for your GPU | 8 | 16.1 GB | Very High | C44 |
F16 | 16 | 30.7 GB | Maximum | F0 |
ollama run starcoder2-15bhuggingface-cli download starcoder2-15bUpgrade options
~$999 MSRP
RTX 3090 24GBBest value~$1,499 MSRP
RTX 4090 24GBBiggest leap~$1,599 MSRP
~$2,499 MSRP