Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
114.1 tok/s
TTFT
1696 ms
Safe context
62K
Memory
17.7 GB / 69.1 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 114.1 tok/s | 2468 ms | 117K |
| Chat | C | Runs well | 114.1 tok/s | 925 ms | 32K |
| Coding | C | Runs well | 114.1 tok/s | 1696 ms | 62K |
| RAG | C | Runs well | 114.1 tok/s | 3084 ms | 117K |
| Reasoning | C | Runs well | 114.1 tok/s | 2005 ms | 62K |
How Meta Llama 3 8B Instruct (8B params) fits at each quantization level on Mac Studio M3 Ultra 96GB (69.1 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 3.1 GB | Low | D30 |
Q3_K_S | 3 | 3.9 GB | Low | D31 |
NVFP4 | 4 | 4.5 GB | Medium | D31 |
Q4_K_M | 4 | 4.9 GB | Medium | D31 |
Q5_K_M | 5 | 5.8 GB | High | D31 |
Q6_K | 6 | 6.6 GB | High | D31 |
Q8_0 | 8 | 8.6 GB | Very High | D32 |
F16Best for your GPU | 16 | 16.4 GB | Maximum | D34 |
