Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
83.5 tok/s
TTFT
2320 ms
Safe context
21K
Memory
12.0 GB / 16.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Tight fit | 83.5 tok/s | 3374 ms | 37K |
| Chat | B | Runs well | 83.5 tok/s | 1265 ms | 12K |
| Coding | B | Runs well | 83.5 tok/s | 2320 ms | 21K |
| RAG | C | Tight fit | 83.5 tok/s | 4218 ms | 37K |
| Reasoning | B | Runs well | 83.5 tok/s | 2742 ms | 21K |
How Gemma 3 12B (12B params) fits at each quantization level on RTX 4080 Super 16GB (16.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 4.7 GB | Low | D35 |
Q3_K_S | 3 | 5.9 GB | Low | D37 |
NVFP4 | 4 | 6.7 GB | Medium | D38 |
Q4_K_M | 4 | 7.3 GB | Medium | D39 |
Q5_K_M | 5 | 8.6 GB | High | C40 |
Q6_KBest for your GPU | 6 | 9.8 GB | High | C42 |
Q8_0 | 8 | 12.8 GB | Very High | C44 |
F16 | 16 | 24.6 GB | Maximum | F0 |
ollama run gemma-3-12bhuggingface-cli download gemma-3-12b