Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
40.2 tok/s
TTFT
5687 ms
Safe context
22K
Memory
210.1 GB / 288.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Tight fit | 40.7 tok/s | 6912 ms | 37K |
| Chat | C | Runs well | 40.7 tok/s | 2592 ms | 12K |
| Coding | C | Runs well | 40.2 tok/s | 5687 ms | 22K |
| RAG | C | Tight fit | 40.7 tok/s | 8640 ms | 37K |
| Reasoning | C | Runs well | 40.7 tok/s | 5616 ms | 22K |
How Baichuan M3 235B (235B params) fits at each quantization level on AMD Instinct MI350X 288GB (288.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 91.7 GB | Low | D36 |
Q3_K_S | 3 | 115.2 GB | Low | D37 |
NVFP4 | 4 |
131.6 GB |
| Medium |
| D39 |
Q4_K_M | 4 | 143.4 GB | Medium | D39 |
Q5_K_M | 5 | 169.2 GB | High | C41 |
Q6_KBest for your GPU | 6 | 192.7 GB | High | C43 |
Q8_0 | 8 | 251.5 GB | Very High | C44 |
F16 | 16 | 481.7 GB | Maximum | F0 |