Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
237.0 tok/s
TTFT
817 ms
Safe context
32K
Memory
63.6 GB / 128.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 230.4 tok/s | 1222 ms | 64K |
| Chat | C | Runs well | 230.4 tok/s | 458 ms | 16K |
| Coding | C | Runs well | 237.0 tok/s | 817 ms | 32K |
| RAG | C | Runs well | 230.4 tok/s | 1528 ms | 64K |
| Reasoning | C | Runs well | 230.4 tok/s | 993 ms | 32K |
How Qwen3-Coder-Next (80B params) fits at each quantization level on AMD Instinct MI300A 128GB (128.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 31.2 GB | Low | D35 |
Q3_K_S | 3 | 39.2 GB | Low | D36 |
NVFP4 | 4 |
ollama run qwen-3-coder-nexthuggingface-cli download qwen-3-coder-next44.8 GB |
| Medium |
| D37 |
Q4_K_M | 4 | 48.8 GB | Medium | D38 |
Q5_K_M | 5 | 57.6 GB | High | D39 |
Q6_K | 6 | 65.6 GB | High | C41 |
Q8_0Best for your GPU | 8 | 85.6 GB | Very High | C44 |
F16 | 16 | 164.0 GB | Maximum | F0 |