Can I run Llama 3.1 8B on an RTX 3060 12GB?
Yes: Llama 3.1 8B needs about 9.88 GB at Q4_K_M with 32K tokens of context, which fits the RTX 3060 12GB with 2.12 GB to spare. At 32K the RTX 3060 holds up to Q5_K_M (10.7 GB), and Q4_K_M runs up to 43K tokens.
Yes Q4_K_M with 32K tokens of context
- Q4_K_M, 32K
- 9.88 GB
- RTX 3060
- 12 GB, 360 GB/s
- To spare
- 2.12 GB
- Tokens/s
- 21–29 tokens/s
At Q4_K_M with 32K tokens of context it writes about 21–29 tokens/s for one request on an RTX 3060 12GB.
Best precision for Llama 3.1 8B on an RTX 3060 12GB
The most precise setting that leaves at least 0.5 GB free; one that fits with less is marked tight.
| Context | Best fit | Memory | Free | Tokens/s |
|---|---|---|---|---|
| 8K | Q8_0 | 10.3 GB | 1.66 GB | 20–28 |
| 32K | Q5_K_M | 10.7 GB | 1.27 GB | 19–27 |
| 128K (full) | Nothing fits | — | — | — |
Llama 3.1 8B on the RTX 3060 as the context fills
One request, FP16 KV cache, 0.5 GB plus 10% overhead; a minus sign is memory missing, and tight is less than 0.5 GB free.
| Context | Q4_K_M | Free | Tokens/s | Q8_0 | Free | Tokens/s |
|---|---|---|---|---|---|---|
| 4K | 6.03 GB | 5.97 GB | 35–49 | 9.79 GB | 2.21 GB | 21–29 |
| 8K | 6.58 GB | 5.42 GB | 32–45 | 10.3 GB | 1.66 GB | 20–28 |
| 16K | 7.68 GB | 4.32 GB | 27–38 | 11.4 GB | 573 MB | 18–25 |
| 32K | 9.88 GB | 2.12 GB | 21–29 | 13.6 GB | −1.64 GB | — |
| 64K | 14.3 GB | −2.28 GB | — | 18.0 GB | −6.04 GB | — |
| 128K (full) | 23.1 GB | −11.1 GB | — | 26.8 GB | −14.8 GB | — |
Llama 3.1 8B on more than one RTX 3060
| Cards | Best at 32K | Tokens/s | Q4_K_M longest | Q4_K_M tokens/s | Rent per hour |
|---|---|---|---|---|---|
| 2× (24 GB) | BF16 | 18–26 | 127K | 38–55 | $0.16 |
| 4× (48 GB) | BF16 | 34–50 | 128K (full) | 66–102 | $0.32 |
Tensor parallel with the combined bandwidth, as in vLLM; llama.cpp splits layers by default and runs at about one card's speed.
Other options
- The next smaller setting, Q3_K_M, takes 8.92 GB at 32K, 3.08 GB under the RTX 3060; it fits with 0.5 GB to spare up to 50K tokens.
- Renting an RTX 3060 12GB costs about $0.08 an hour (median on getdeploying.com, 2026-09-29): $0.76–1.1 per million tokens at 21–29 tokens/s.
- Smallest setup for Q4_K_M at 32K: RTX 3060 12GB.
Run Llama 3.1 8B on the RTX 3060 with llama-server
llama-server -hf bartowski/Meta-Llama-3.1-8B-Instruct-GGUF:Q4_K_M -c 44032Meta-Llama-3.1-8B-Instruct-Q4_K_M.gguf, 4.9 GB, from bartowski/
Questions
Can I run Llama 3.1 8B on an RTX 3060 12GB?
Yes: Llama 3.1 8B needs about 9.88 GB at Q4_K_M with 32K tokens of context, which fits the RTX 3060 12GB with 2.12 GB to spare. At 32K the RTX 3060 holds up to Q5_K_M (10.7 GB), and Q4_K_M runs up to 43K tokens.
How fast is Llama 3.1 8B on an RTX 3060 12GB?
At Q4_K_M with 32K tokens of context it writes about 21–29 tokens/s for one request on an RTX 3060 12GB.
What does a second RTX 3060 12GB change for Llama 3.1 8B?
Two RTX 3060 12GB cards (24 GB in one tensor-parallel group) hold Llama 3.1 8B at BF16 with 32K, and Q4_K_M up to 127K tokens, at about 38–55 tokens/s.
Try other settings in the VRAM calculator or the speed calculator. See also Llama 3.1 8B VRAM requirements, what LLMs an RTX 3060 12GB can run and every pair, or detect your own GPU. Model data checked .