Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
143.4 tok/s
TTFT
1350 ms
Safe context
60K
Memory
12.9 GB / 48.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 143.4 tok/s | 1964 ms | 107K |
| Chat | C | Runs well | 143.4 tok/s | 736 ms | 31K |
| Coding | C | Runs well | 143.4 tok/s | 1350 ms | 60K |
| RAG | C | Runs well | 143.4 tok/s | 2455 ms | 107K |
| Reasoning | C | Runs well | 143.4 tok/s | 1596 ms | 60K |
How Qwen3.5 9B (9B params) fits at each quantization level on RTX 6000 Ada 48GB (48.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 3.5 GB | Low | D31 |
Q3_K_S | 3 | 4.4 GB | Low | D31 |
NVFP4 | 4 | 5.0 GB | Medium | D32 |
Q4_K_M | 4 | 5.5 GB | Medium | D32 |
Q5_K_M | 5 | 6.5 GB | High | D32 |
Q6_K | 6 | 7.4 GB | High | D33 |
Q8_0 | 8 | 9.6 GB | Very High | D34 |
F16Best for your GPU | 16 | 18.5 GB | Maximum | D38 |
huggingface-cli download hf-lmstudio-community--qwen3-5-9b-gguf