RX 7900 XT 20GBBudget pick
C307.3 tok/s decode
~$899 MSRP
Can it run?
Runs well
Using Q6_K in Ollama
Fit status
Runs well
Decode
43.5 tok/s
TTFT
4447 ms
Safe context
44K
Memory
6.2 GB / 17.3 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 43.5 tok/s | 6468 ms | 89K |
| Chat | C | Runs well | 43.5 tok/s | 2426 ms | 22K |
| Coding | C | Runs well | 43.5 tok/s | 4447 ms | 44K |
| RAG | C | Runs well | 43.5 tok/s | 8085 ms | 89K |
| Reasoning | C | Runs well | 43.5 tok/s | 5255 ms | 44K |
How gemma 2 2b it (2B params) fits at each quantization level on MacBook Pro M3 24GB (17.3 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 0.8 GB | Low | D31 |
Q3_K_S | 3 | 1.0 GB | Low | D31 |
NVFP4 | 4 | 1.1 GB | Medium | D31 |
Q4_K_M | 4 | 1.2 GB | Medium | D31 |
Q5_K_M | 5 | 1.4 GB | High | D31 |
Q6_K | 6 | 1.6 GB | High | D32 |
Q8_0 | 8 | 2.1 GB | Very High | D32 |
F16Best for your GPU | 16 | 4.1 GB | Maximum | D35 |
huggingface-cli download hf-bartowski--gemma-2-2b-it-ggufUpgrade options
~$899 MSRP
~$999 MSRP
~$1,999 MSRP
~$2,499 MSRP