Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
68.0 tok/s
TTFT
2848 ms
Safe context
25K
Memory
25.9 GB / 40.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | B | Runs well | 68.0 tok/s | 4143 ms | 43K |
| Chat | C | Runs well | 68.0 tok/s | 1553 ms | 13K |
| Coding | B | Runs well | 68.0 tok/s | 2848 ms | 25K |
| RAG | B | Runs well | 68.0 tok/s | 5178 ms | 43K |
| Reasoning | B | Runs well | 68.0 tok/s | 3366 ms | 25K |
How Qwen3.5 27B (27B params) fits at each quantization level on NVIDIA A100 40GB (40.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 10.5 GB | Low | D35 |
Q3_K_S | 3 | 13.2 GB | Low | D37 |
NVFP4 | 4 |
huggingface-cli download hf-unsloth--qwen3-5-27b-gguf15.1 GB |
| Medium |
| D38 |
Q4_K_M | 4 | 16.5 GB | Medium | D38 |
Q5_K_M | 5 | 19.4 GB | High | D40 |
Q6_K | 6 | 22.1 GB | High | C41 |
Q8_0Best for your GPU | 8 | 28.9 GB | Very High | C45 |
F16 | 16 | 55.4 GB | Maximum | F0 |