Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
78.5 tok/s
TTFT
2465 ms
Safe context
53K
Memory
29.2 GB / 96.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 91.6 tok/s | 3073 ms | 93K |
| Chat | C | Runs well | 91.6 tok/s | 1152 ms | 28K |
| Coding | C | Runs well | 78.5 tok/s | 2465 ms | 53K |
| RAG | C | Runs well | 91.6 tok/s | 3841 ms | 93K |
| Reasoning | C | Runs well | 91.6 tok/s | 2497 ms | 53K |
How Devstral Small 2 24B Instruct (24B params) fits at each quantization level on RTX PRO 6000 Blackwell Server Edition 96GB (96.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 9.4 GB | Low | D32 |
Q3_K_S | 3 | 11.8 GB | Low | D32 |
NVFP4 | 4 |
ollama run devstral-small-2-24bhuggingface-cli download devstral-small-2-24b13.4 GB |
| Medium |
| D32 |
Q4_K_M | 4 | 14.6 GB | Medium | D33 |
Q5_K_M | 5 | 17.3 GB | High | D33 |
Q6_K | 6 | 19.7 GB | High | D34 |
Q8_0 | 8 | 25.7 GB | Very High | D35 |
F16Best for your GPU | 16 | 49.2 GB | Maximum | C40 |