Llama 2-13B
Llama 2-13B is an open-weights language model from Meta, released July 18, 2023. It has 13B parameters (dense). On the Epoch Capabilities Index it scores 96.7, rank 221 of 256 rated models.
Check it against your own PC →
Can you run Llama 2-13B locally?
The smallest common GPU it runs on is 6 GB of VRAM (with 32 GB of RAM): IQ4_XS, split GPU / CPU, roughly 5 tokens per second. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Runs, slowly | IQ4_XS (good) | 5.2 GB | 4.8 GB | ~5 |
| 8 GB VRAM | Runs, slowly | Q4_K_M (very good) | 7.2 GB | 3.9 GB | ~6 |
| 12 GB VRAM | Runs on GPU | Q4_K_M (very good) | 11.1 GB | – | ~26 |
| 16 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 13.8 GB | – | ~27 |
| 24 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 13.8 GB | – | ~50 |
| 32 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 13.8 GB | – | ~92 |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| MMLU | 40.8% |
Links
Models with a similar score
DeepSeek-Coder-V2-Lite-Base · Falcon 2 11B · Nemotron-4 15B · internlm-20b · PaLM 2-M · Qwen2.5-Coder-3B · LLaMA-33B · Phi-2
More from Meta
Muse Ember · Muse Spark Open · Muse Wonder · Muse Wonder Full · Muse Spark 1.3 · Muse Spark 1.3 Contributor · Muse Glimmer · Muse Glimmer 30B assistant