Can it run?
Runs well
Using Q6_K in Ollama
Fit status
Runs well
Decode
220.7 tok/s
TTFT
877 ms
Safe context
76K
Memory
9.7 GB / 46.1 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 220.7 tok/s | 1276 ms | 152K |
| Chat | C | Runs well | 220.7 tok/s | 478 ms | 38K |
| Coding | C | Runs well | 220.7 tok/s | 877 ms | 76K |
| RAG | C | Runs well | 220.7 tok/s | 1595 ms | 152K |
| Reasoning | C | Runs well | 220.7 tok/s | 1037 ms | 76K |
How Llama 3.2 1B Instruct Q8 0 (1B params) fits at each quantization level on MacBook Pro M4 Pro 64GB (46.1 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 0.4 GB | Low | D30 |
Q3_K_S | 3 | 0.5 GB | Low | D30 |
NVFP4 | 4 | 0.6 GB | Medium | D30 |
Q4_K_M | 4 | 0.6 GB | Medium | D30 |
Q5_K_M | 5 | 0.7 GB | High | D30 |
Q6_K | 6 | 0.8 GB | High | D30 |
Q8_0 | 8 | 1.1 GB | Very High | D30 |
F16Best for your GPU | 16 | 2.1 GB | Maximum | D30 |
huggingface-cli download hf-hugging-quants--llama-3-2-1b-instruct-q8-0-gguf