Can it run?
Runs well
Using Q4_K_M in vLLM
Fit status
Runs well
Decode
148.7 tok/s
TTFT
1302 ms
Safe context
23K
Memory
88.8 GB / 128.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | B | Runs well | 148.7 tok/s | 1894 ms | 46K |
| Chat | B | Runs well | 148.7 tok/s | 710 ms | 12K |
| Coding | B | Runs well | 148.7 tok/s | 1302 ms | 23K |
| RAG | B | Runs well | 148.7 tok/s | 2367 ms | 46K |
| Reasoning | B | Runs well | 148.7 tok/s | 1539 ms | 23K |
How Mistral Small 4 119B (119B params) fits at each quantization level on AMD Instinct MI300A 128GB (128.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 46.4 GB | Low | D37 |
Q3_K_S | 3 | 58.3 GB | Low | D39 |
NVFP4 | 4 |
huggingface-cli download mistral-small-4-119b66.6 GB |
| Medium |
| C41 |
Q4_K_M | 4 | 72.6 GB | Medium | C42 |
Q5_K_M | 5 | 85.7 GB | High | C44 |
Q6_KBest for your GPU | 6 | 97.6 GB | High | C45 |
Q8_0 | 8 | 127.3 GB | Very High | C45 |
F16 | 16 | 244.0 GB | Maximum | F0 |