Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
132.2 tok/s
TTFT
1464 ms
Safe context
37K
Memory
10.5 GB / 24.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 130.4 tok/s | 2160 ms | 65K |
| Chat | C | Runs well | 130.4 tok/s | 810 ms | 19K |
| Coding | C | Runs well | 132.2 tok/s | 1464 ms | 37K |
| RAG | C | Runs well | 130.4 tok/s | 2700 ms | 65K |
| Reasoning | C | Runs well | 130.4 tok/s | 1755 ms | 37K |
How Qwen3.5 9B (9B params) fits at each quantization level on RTX 3090 Ti 24GB (24.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 3.5 GB | Low | D33 |
Q3_K_S | 3 | 4.4 GB | Low | D33 |
NVFP4 | 4 |
huggingface-cli download hf-unsloth--qwen3-5-9b-gguf5.0 GB |
| Medium |
| D34 |
Q4_K_M | 4 | 5.5 GB | Medium | D34 |
Q5_K_M | 5 | 6.5 GB | High | D35 |
Q6_K | 6 | 7.4 GB | High | D36 |
Q8_0 | 8 | 9.6 GB | Very High | D38 |
F16Best for your GPU | 16 | 18.5 GB | Maximum | C45 |