Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
268.3 tok/s
TTFT
722 ms
Safe context
54K
Memory
76.4 GB / 256.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 268.3 tok/s | 1050 ms | 107K |
| Chat | C | Runs well | 268.3 tok/s | 394 ms | 27K |
| Coding | C | Runs well | 268.3 tok/s | 722 ms | 54K |
| RAG | C | Runs well | 268.3 tok/s | 1312 ms | 107K |
| Reasoning | C | Runs well | 268.3 tok/s | 853 ms | 54K |
How Qwen3-Coder-Next (80B params) fits at each quantization level on AMD Instinct MI325X 256GB (256.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 31.2 GB | Low | D32 |
Q3_K_S | 3 | 39.2 GB | Low | D33 |
NVFP4 | 4 |
ollama run qwen-3-coder-nexthuggingface-cli download qwen-3-coder-next44.8 GB |
| Medium |
| D33 |
Q4_K_M | 4 | 48.8 GB | Medium | D34 |
Q5_K_M | 5 | 57.6 GB | High | D34 |
Q6_K | 6 | 65.6 GB | High | D35 |
Q8_0 | 8 | 85.6 GB | Very High | D37 |
F16Best for your GPU | 16 | 164.0 GB | Maximum | C43 |