Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
89.2 tok/s
TTFT
2170 ms
Safe context
27K
Memory
23.6 GB / 40.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | B | Runs well | 89.2 tok/s | 3156 ms | 47K |
| Chat | C | Runs well | 89.2 tok/s | 1184 ms | 15K |
| Coding | C | Runs well | 89.2 tok/s | 2170 ms | 27K |
| RAG | B | Runs well | 89.2 tok/s | 3945 ms | 47K |
| Reasoning | C | Runs well | 89.2 tok/s | 2564 ms | 27K |
How Magistral Small 2507 (24B params) fits at each quantization level on NVIDIA A100 40GB (40.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 9.4 GB | Low | D34 |
Q3_K_S | 3 | 11.8 GB | Low | D36 |
NVFP4 | 4 | 13.4 GB | Medium | D36 |
Q4_K_M | 4 | 14.6 GB | Medium | D37 |
Q5_K_M | 5 | 17.3 GB | High | D39 |
Q6_K | 6 | 19.7 GB | High | D40 |
Q8_0Best for your GPU | 8 | 25.7 GB | Very High | C43 |
F16 | 16 | 49.2 GB | Maximum | F0 |
ollama run magistral-small-2507huggingface-cli download magistral-small-2507