Can it run?
Runs well
Using Q4_K_M in vLLM
Fit status
Runs well
Decode
88.8 tok/s
TTFT
2179 ms
Safe context
25K
Memory
115.4 GB / 180.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | B | Runs well | 88.8 tok/s | 3170 ms | 43K |
| Chat | C | Runs well | 88.8 tok/s | 1189 ms | 14K |
| Coding | B | Runs well | 88.8 tok/s | 2179 ms | 25K |
| RAG | B | Runs well | 88.8 tok/s | 3962 ms | 43K |
| Reasoning | B | Runs well | 88.8 tok/s | 2575 ms | 25K |
How Pixtral Large 124B (124B params) fits at each quantization level on NVIDIA B200 180GB (180.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 48.4 GB | Low | D35 |
Q3_K_S | 3 | 60.8 GB | Low | D37 |
NVFP4 | 4 | 69.4 GB | Medium | D38 |
Q4_K_M | 4 | 75.6 GB | Medium | D38 |
Q5_K_M | 5 | 89.3 GB | High | C40 |
Q6_K | 6 | 101.7 GB | High | C41 |
Q8_0Best for your GPU | 8 | 132.7 GB | Very High | C44 |
F16 | 16 | 254.2 GB | Maximum | F0 |
huggingface-cli download pixtral-large-124b