Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
69.8 tok/s
TTFT
2773 ms
Safe context
49K
Memory
31.5 GB / 96.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 81.4 tok/s | 3457 ms | 86K |
| Chat | C | Runs well | 81.4 tok/s | 1297 ms | 26K |
| Coding | C | Runs well | 69.8 tok/s | 2773 ms | 49K |
| RAG | C | Runs well | 81.4 tok/s | 4322 ms | 86K |
| Reasoning | C | Runs well | 81.4 tok/s | 2809 ms | 49K |
How Qwen3.5 27B (27B params) fits at each quantization level on RTX PRO 6000 Blackwell Server Edition 96GB (96.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 10.5 GB | Low | D32 |
Q3_K_S | 3 | 13.2 GB | Low | D32 |
NVFP4 | 4 |
huggingface-cli download hf-unsloth--qwen3-5-27b-gguf15.1 GB |
| Medium |
| D33 |
Q4_K_M | 4 | 16.5 GB | Medium | D33 |
Q5_K_M | 5 | 19.4 GB | High | D34 |
Q6_K | 6 | 22.1 GB | High | D34 |
Q8_0 | 8 | 28.9 GB | Very High | D36 |
F16Best for your GPU | 16 | 55.4 GB | Maximum | C42 |