Llama 2 models
8 Llama 2 models from Meta on Hugging Face, from 6.7B to 13.0B parameters, published by Meta in F16, with a 4K-token context. The smallest official model, Llama-2-7b-chat-hf, needs about 15.1 GB of VRAM at its published precision; the cheapest live fit is V100 at $0.110/hr.
Part of the Llama series · Previous generation: Llama 1 · Next generation: Llama 3
Pick a size
One row per official Llama 2 size: the VRAM it needs at each precision, the cheapest GPU that holds it today, and the KV cache for a 32K-token context.
| Model | Parameters | Native VRAM | FP8 VRAM | INT4 VRAM | Live GPU fit (native) | Est. $/hr | KV cache at 32K |
|---|---|---|---|---|---|---|---|
| Llama-2-7b-hf | 6.7B | 15.1 GB | 7.5 GB | 3.8 GB | V100 | $0.110/hr | 16.0 GB |
| Llama-2-13b-chat-hf | 13.0B | 29.1 GB | 14.5 GB | 7.3 GB | V100 | $0.220/hr | 25.0 GB |
VRAM is weights times a flat 1.2 overhead; the KV cache is a separate per-model figure at 16-bit, shown where the architecture is published. See the methodology.
Official models (4)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Llama-2-7b-chat-hf | 6.7B | F16 | 15.1 GB | V100 | $0.110/hr |
| Llama-2-7b-hf | 6.7B | F16 | 15.1 GB | V100 | $0.110/hr |
| Llama-2-13b-chat-hf | 13.0B | F16 | 29.1 GB | V100 | $0.220/hr |
| Llama-2-13b-hf | 13.0B | F16 | 29.1 GB | V100 | $0.220/hr |
Fine-tunes and community models (4)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| llama-2-7b | 6.7B | F16 | 15.1 GB | V100 | $0.110/hr |
| llama-2-7b-chat | 6.7B | F16 | 15.1 GB | V100 | $0.110/hr |
| HarmBench-Llama-2-13b-cls | 13.0B | BF16 | 29.1 GB | RTX A6000 | $0.363/hr |
| llama2-13b-chat-hf | 13.0B | F16 | 29.1 GB | V100 | $0.220/hr |
VRAM is for the precision the model is published in: the weights times a flat 1.2 overhead, with the KV cache not included. See the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
Run and fine-tune Llama 2
- How to run Llama 2: engine support, chat template and context flags, with a serving command sized per model.
- How to fine-tune Llama 2: LoRA and QLoRA memory per model, and the license limits on derivatives.
More from Meta
- Llama 3.2 (15 models)
- Llama 3.1 (40 models)
- Llama 3 (19 models)
- Llama 3.3 (4 models)
- Code Llama (3 models)
- Llama Guard (4 models)
- Llama 4 (2 models)
- Llama 1 (1 models)