Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
224.3 tok/s
TTFT
863 ms
Safe context
29K
Memory
5.4 GB / 10.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 236.7 tok/s | 1190 ms | 54K |
| Chat | C | Runs well | 236.7 tok/s | 446 ms | 15K |
| Coding | C | Runs well | 224.3 tok/s | 863 ms | 29K |
| RAG | C | Runs well | 236.7 tok/s | 1487 ms | 54K |
| Reasoning | C | Runs well | 236.7 tok/s | 967 ms | 29K |
How Qwen3.5 4B (4B params) fits at each quantization level on RTX 3080 10GB (10.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 1.6 GB | Low | D33 |
Q3_K_S | 3 | 2.0 GB | Low | D34 |
NVFP4 | 4 |
| Medium |
| D34 |
Q4_K_M | 4 | 2.4 GB | Medium | D35 |
Q5_K_M | 5 | 2.9 GB | High | D36 |
Q6_K | 6 | 3.3 GB | High | D37 |
Q8_0Best for your GPU | 8 | 4.3 GB | Very High | D39 |
F16 | 16 | 8.2 GB | Maximum | C44 |