LLaMA-65B
LLaMA-65B is an open-weights language model from Meta, released February 24, 2023. It has 65B parameters (dense). On the Epoch Capabilities Index it scores 101.7, rank 214 of 256 rated models.
Check it against your own PC →
Can you run LLaMA-65B locally?
The smallest common GPU it runs on is 32 GB of VRAM (with 32 GB of RAM): IQ3_XXS, fully on GPU, roughly 41 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 | Very slow | IQ3_XXS (noticeable loss) | 5.2 GB | 24.6 GB | ~1 |
| 8 GB VRAM | Very slow | IQ3_XXS (noticeable loss) | 7.2 GB | 22.6 GB | ~1 |
| 12 GB VRAM | Very slow | IQ3_XXS (noticeable loss) | 11.2 GB | 18.6 GB | ~2 |
| 16 GB VRAM | Very slow | IQ3_XXS (noticeable loss) | 15.2 GB | 14.6 GB | ~2 |
| 24 GB VRAM | Very slow | IQ3_XXS (noticeable loss) | 23.2 GB | 6.6 GB | ~4 |
| 32 GB VRAM | Runs on GPU | IQ3_XXS (noticeable loss) | 29.8 GB | – | ~41 |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| MMLU | 51.2% |
Links
Models with a similar score
Gemma 7B · Mistral 7B v0.1 · Yi-9B · Qwen-14B · DeepSeek-Coder-V2-Lite-Base · Falcon 2 11B · Nemotron-4 15B · internlm-20b
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