Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
138.3 tok/s
TTFT
1400 ms
Safe context
23K
Memory
8.5 GB / 12.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | B | Runs well | 138.3 tok/s | 2036 ms | 39K |
| Chat | B | Runs well | 138.3 tok/s | 764 ms | 12K |
| Coding | B | Runs well | 138.3 tok/s | 1400 ms | 23K |
| RAG | B | Runs well | 138.3 tok/s | 2545 ms | 39K |
| Reasoning | B | Runs well | 138.3 tok/s | 1654 ms | 23K |
How Llama 3.1 8B (8B params) fits at each quantization level on RTX 3080 Ti 12GB (12.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 3.1 GB | Low | D34 |
Q3_K_S | 3 | 3.9 GB | Low | D35 |
NVFP4 | 4 | 4.5 GB | Medium | D36 |
Q4_K_M | 4 | 4.9 GB | Medium | D37 |
Q5_K_M | 5 | 5.8 GB | High | D39 |
Q6_K | 6 | 6.6 GB | High | C40 |
Q8_0Best for your GPU | 8 | 8.6 GB | Very High | C43 |
F16 | 16 | 16.4 GB | Maximum | F0 |
ollama run llama-3.1-8bhuggingface-cli download llama-3.1-8b