C68.5 tok/s decode
Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
31.6 tok/s
TTFT
6132 ms
Safe context
26K
Memory
28.8 GB / 46.1 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 28.2 tok/s | 9996 ms | 45K |
| Chat | C | Runs well | 28.2 tok/s | 3748 ms | 14K |
| Coding | C | Runs well | 31.6 tok/s | 6132 ms | 26K |
| RAG | C | Runs well | 28.2 tok/s | 12494 ms | 45K |
| Reasoning | C | Runs well | 28.2 tok/s | 8121 ms | 26K |
How gemma 3 27b it (27B params) fits at each quantization level on Mac Studio M2 Ultra 64GB (46.1 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 10.5 GB | Low | D34 |
Q3_K_S | 3 | 13.2 GB | Low | D36 |
NVFP4 | 4 |
Upgrade options
15.1 GB |
| Medium |
| D36 |
Q4_K_M | 4 | 16.5 GB | Medium | D37 |
Q5_K_M | 5 | 19.4 GB | High | D38 |
Q6_K | 6 | 22.1 GB | High | D40 |
Q8_0Best for your GPU | 8 | 28.9 GB | Very High | C43 |
F16 | 16 | 55.4 GB | Maximum | F0 |