NVIDIA B200 180GBBudget pickB89.6 tok/s decode
~$30,000 MSRP
Can it run?
Runs well
Using Q4_K_M in vLLM
Fit status
Runs well
Decode
46.1 tok/s
TTFT
4203 ms
Safe context
20K
Memory
110.7 GB / 141.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Tight fit | 46.1 tok/s | 6114 ms | 35K |
| Chat | B | Runs well | 46.1 tok/s | 2293 ms | 11K |
| Coding | B | Runs well | 46.1 tok/s | 4203 ms | 20K |
| RAG | C | Tight fit | 46.1 tok/s | 7642 ms | 35K |
| Reasoning | B | Runs well | 46.1 tok/s | 4967 ms | 20K |
How Devstral 2 123B Instruct (123B params) fits at each quantization level on NVIDIA H200 PCIe 141GB (141.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 48.0 GB | Low | D37 |
Q3_K_S | 3 | 60.3 GB | Low | D39 |
NVFP4 | 4 |
huggingface-cli download devstral-2-123bUpgrade options
NVIDIA B200 180GBBudget pick~$30,000 MSRP
68.9 GB |
| Medium |
| C40 |
Q4_K_M | 4 | 75.0 GB | Medium | C41 |
Q5_K_M | 5 | 88.6 GB | High | C43 |
Q6_KBest for your GPU | 6 | 100.9 GB | High | C45 |
Q8_0 | 8 | 131.6 GB | Very High | C45 |
F16 | 16 | 252.2 GB | Maximum | F0 |