Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
47.2 tok/s
TTFT
4101 ms
Safe context
31K
Memory
24.4 GB / 48.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 53.8 tok/s | 5237 ms | 55K |
| Chat | C | Runs well | 53.8 tok/s | 1964 ms | 17K |
| Coding | C | Runs well | 47.2 tok/s | 4101 ms | 31K |
| RAG | C | Runs well | 53.8 tok/s | 6546 ms | 55K |
| Reasoning | C | Runs well | 53.8 tok/s | 4255 ms | 31K |
How Devstral Small 2 24B Instruct (24B params) fits at each quantization level on RTX 6000 Ada 48GB (48.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 9.4 GB | Low | D34 |
Q3_K_S | 3 | 11.8 GB | Low | D35 |
NVFP4 | 4 |
ollama run devstral-small-2-24bhuggingface-cli download devstral-small-2-24b13.4 GB |
| Medium |
| D35 |
Q4_K_M | 4 | 14.6 GB | Medium | D36 |
Q5_K_M | 5 | 17.3 GB | High | D37 |
Q6_K | 6 | 19.7 GB | High | D38 |
Q8_0Best for your GPU | 8 | 25.7 GB | Very High | C41 |
F16 | 16 | 49.2 GB | Maximum | F0 |