Can it run?
Runs well
Using Q4_K_M in vLLM
Fit status
Runs well
Decode
60.2 tok/s
TTFT
3219 ms
Safe context
22K
Memory
94.9 GB / 128.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Tight fit | 58.5 tok/s | 4815 ms | 37K |
| Chat | B | Runs well | 58.5 tok/s | 1806 ms | 12K |
| Coding | B | Runs well | 60.2 tok/s | 3219 ms | 22K |
| RAG | C | Tight fit | 58.5 tok/s | 6019 ms | 37K |
| Reasoning | B | Runs well | 58.5 tok/s | 3912 ms | 22K |
How Command R+ 104B (104B params) fits at each quantization level on AMD Instinct MI300A 128GB (128.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 40.6 GB | Low | D36 |
Q3_K_S | 3 | 51.0 GB | Low | D38 |
NVFP4 | 4 |
ollama run command-r-plus-104bhuggingface-cli download command-r-plus-104b58.2 GB |
| Medium |
| D39 |
Q4_K_M | 4 | 63.4 GB | Medium | D40 |
Q5_K_M | 5 | 74.9 GB | High | C42 |
Q6_KBest for your GPU | 6 | 85.3 GB | High | C44 |
Q8_0 | 8 | 111.3 GB | Very High | C44 |
F16 | 16 | 213.2 GB | Maximum | F0 |