How much VRAM does Code Llama 13B need?
Code Llama 13B needs about 8.5 GB of VRAM at Q4_K_M, or 14.3 GB at Q8_0 and 26.2 GB at FP16. That fits a consumer card: the smallest in our set that holds it at Q4_K_M is the RTX 3060 12GB (12 GB). Weights plus runtime overhead; KV cache for your context length is on top.
Weights-plus-runtime footprint across the six most common quantizations. KV cache is context-dependent and comes on top. A full 16K window is a real share of the budget; size it precisely in the GPU Memory Calculator.
This is the base checkpoint: pretrained on raw text and not tuned to follow instructions, so it continues text rather than answering prompts. That makes it the right starting point if you plan to fine-tune your own model. For chat or assistant work, use Code Llama 13B Instruct — same parameter count, so the memory figures below are identical.
Can your GPU run Code Llama 13B?
Pick your card — the answer below is computed by the same engine that produces every figure on this page.
RTX 3060 12GB runs Code Llama 13B at Q4_K_M with 2.9 GB to spare; room left for context and everyday runtime pressure.
Weights plus runtime overhead. KV cache is not included — it depends on context length and this model’s published architecture.
| Weights (Q4_K_M) | 7.3 GB |
| Runtime overhead | 1.1 GB |
| Estimated total | 8.5 GB |
| Usable VRAM (95% of 12 GB) | 11.4 GB |
Across every GPU we track
- RTX 3060 12GBRecommended+2.9 GB
- RTX 4060 Ti 16GBExcellent+6.7 GB
- RTX 4070 Ti SUPERExcellent+6.7 GB
- RTX 3090Excellent+14.3 GB
- RTX 4090Excellent+14.3 GB
- RTX 5090Excellent+21.9 GB
- A100 40GBExcellent+29.5 GB
- RTX A6000Excellent+37.1 GB
- L40SExcellent+37.1 GB
- A100 80GBExcellent+67.5 GB
- H100 80GBExcellent+67.5 GB
- H200 141GBExcellent+125.5 GB
Estimates, not guarantees: real usage moves with runtime, driver, batch size and context. Size a specific context window in the GPU Memory Calculator.
How was this number calculated? Every figure here comes from Bitpute’s documented calculation methodology — parameters, precision, quantization, runtime overhead and usable VRAM, each formula written out in full.
Memory by quantization
| Quant | Weights | + runtime overhead |
|---|---|---|
| FP16 | 24.2 GB | 26.2 GB |
| Q8_0 | 12.9 GB | 14.3 GB |
| Q6_K | 9.9 GB | 11.2 GB |
| Q5_K_M | 8.6 GB | 9.8 GB |
| Q4_K_M | 7.3 GB | 8.5 GB |
| Q4_0 | 6.8 GB | 7.9 GB |
Overhead = 0.75 GB + 5% of weights (CUDA context, buffers). Add KV cache on top. A full 16K window is a real share of the budget.
Which quantization should you actually run?
On a mainstream 12 GB card, Q5_K_M (9.8 GB) is the sweet spot for Code Llama 13B — near-lossless and though context headroom gets tight. For Q8_0 (14.3 GB) step up to RTX 4090; full FP16 (26.2 GB) needs RTX A6000 and is rarely worth it at this size. For a coding assistant, prefer Q5_K_M or higher: aggressive quantization tends to show up as subtle syntax and logic slips.
Single-GPU fit at Q4_K_M (8.5 GB + KV)
RTX 3060 12GBRTX 4060 Ti 16GBRTX 4070 Ti SUPERRTX 3090RTX 4090RTX 5090A100 40GBRTX A6000L40SA100 80GBH100 80GBH200 141GB
Where the memory goes
Weights and overhead are exact for Q4_K_M. The KV bar assumes a generic 32-layer transformer at 8K context — your model’s layer count and attention scheme move it, which is what the calculator is for.
Weighing the cheapest card that fits against the next one up? RTX 3060 12GB vs RTX 4060 Ti 16GB compares them on bandwidth, power and which models each one holds.
Common questions
How much VRAM does Code Llama 13B need?
Code Llama 13B needs about 8.5 GB of VRAM at Q4_K_M, 14.3 GB at Q8_0, or 26.2 GB at FP16. That is weights plus about 0.75 GB of runtime overhead and 5% of weight size; KV cache is additional and depends on context length.
What GPU can run Code Llama 13B?
At Q4_K_M the smallest card in our set that fits is the RTX 3060 12GB (12 GB usable at 95%). At FP16 you need the RTX 5090 (32 GB) or larger.
Can Code Llama 13B run on 24 GB?
Yes. At Q4_K_M it needs about 8.5 GB, which fits inside the ~22.8 GB usable on a 24 GB card.
Same memory footprint as 2 other checkpoints
Code Llama 13B has an identical parameter count to these, so every figure on this page applies to them unchanged. What varies is different checkpoint roles (Python-specialised, instruction-tuned) — not the size. Choose on capability and licence; the memory budget is identical.
Code Llama 13B InstructCode Llama 13B Python
More in the Code Llama family
Code Llama 7BCode Llama 7B InstructCode Llama 7B PythonCode Llama 13B InstructCode Llama 13B PythonCode Llama 34B