Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
85.4 tok/s
TTFT
2266 ms
Safe context
26K
Memory
14.9 GB / 24.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | B | Runs well | 85.4 tok/s | 3296 ms | 33K |
| Chat | C | Runs well | 85.4 tok/s | 1236 ms | 14K |
| Coding | B | Runs well | 85.4 tok/s | 2266 ms | 26K |
| RAG | B | Runs well | 85.4 tok/s | 4120 ms | 33K |
| Reasoning | B | Runs well | 85.4 tok/s | 2678 ms | 26K |
How Phi-4-reasoning-plus 14B (14.699999809265137B params) fits at each quantization level on RTX 4090 24GB (24.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 5.7 GB | Low | D34 |
Q3_K_S | 3 | 7.2 GB | Low | D36 |
NVFP4 | 4 |
ollama run phi-4-reasoning-plus-14bhuggingface-cli download phi-4-reasoning-plus-14b| Medium |
| D37 |
Q4_K_M | 4 | 9.0 GB | Medium | D37 |
Q5_K_M | 5 | 10.6 GB | High | D39 |
Q6_K | 6 | 12.1 GB | High | C40 |
Q8_0Best for your GPU | 8 | 15.7 GB | Very High | C43 |
F16 | 16 | 30.1 GB | Maximum | F0 |