Can it run?
Runs well
Using Q4_K_M in vLLM
Fit status
Runs well
Decode
153.0 tok/s
TTFT
1266 ms
Safe context
32K
Memory
95.2 GB / 192.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 165.7 tok/s | 1699 ms | 64K |
| Chat | C | Runs well | 165.7 tok/s | 637 ms | 16K |
| Coding | C | Runs well | 153.0 tok/s | 1266 ms | 32K |
| RAG | C | Runs well | 165.7 tok/s | 2124 ms | 64K |
| Reasoning | C | Runs well | 165.7 tok/s | 1381 ms | 32K |
How Mistral Small 4 119B (119B params) fits at each quantization level on AMD Instinct MI300X 192GB (192.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 46.4 GB | Low | D35 |
Q3_K_S | 3 | 58.3 GB | Low | D36 |
NVFP4 | 4 |
huggingface-cli download mistral-small-4-119b66.6 GB |
| Medium |
| D37 |
Q4_K_M | 4 | 72.6 GB | Medium | D38 |
Q5_K_M | 5 | 85.7 GB | High | D39 |
Q6_K | 6 | 97.6 GB | High | C40 |
Q8_0Best for your GPU | 8 | 127.3 GB | Very High | C44 |
F16 | 16 | 244.0 GB | Maximum | F0 |