Llama3.1 60B needs about 40.1 GB of memory at Q4_K_M (a 37.3 GB file) with an 8K context window, so it fits on a 48 GB GPU. This repository has 12 quants, from 23.2 GB (Q2_K) to 43.9 GB (Q5_K_M).
Llama3.1 60B is a 62B-parameter Llama model in GGUF format, quantized by mradermacher. Its trained context length is 8K tokens. The repository has 12 quant files, from 23.2 GB (Q2_K) to 43.9 GB (Q5_K_M).
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 23.2 GB | 26.0 GB | Too big | Get file → | |
| IQ3_XS | 25.7 GB | 28.6 GB | Too big | Get file → | |
| Q3_K_S | 27.1 GB | 29.9 GB | Too big | Get file → | |
| IQ3_S | 27.1 GB | 30.0 GB | Too big | Get file → | |
| IQ3_M | 28.0 GB | 30.9 GB | Too big | Get file → | |
| Q3_K_M | 30.1 GB | 32.9 GB | Too big | Get file → | |
| Q3_K_L | 32.7 GB | 35.5 GB | Too big | Get file → | |
| IQ4_XS | 33.6 GB | 36.5 GB | Too big | Get file → | |
| Q4_K_S | 35.4 GB | 38.2 GB | Too big | Get file → | |
| Q4_K_M | 37.3 GB | 40.1 GB | Too big | Get file → | |
| Q5_K_S | 42.8 GB | 45.6 GB | Too big | Get file → | |
| Q5_K_M | 43.9 GB | 46.7 GB | Too big | Get file → |
What GPU do you need for Llama3.1 60B?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: no quant fits at an 8K context.
- 32 GB VRAM: the best quant that fits is IQ3_M (28.0 GB file, about 30.9 GB in use).
- 48 GB VRAM: the best quant that fits is Q5_K_M (43.9 GB file, about 46.7 GB in use).
Which Llama3.1 60B quant should you choose?
Q4_K_M is the usual starting point for Llama3.1 60B: it balances file size and quality well. Here it is a 37.3 GB file.
If memory is tight, the smallest file is Q2_K at 23.2 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Llama3.1 60B
How much VRAM does Llama3.1 60B need?
About 40.1 GB at Q4_K_M with an 8K context window. That is the 37.3 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Llama3.1 60B on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 26.0 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Llama3.1 60B GGUF?
It depends on the quant: Q2_K is 23.2 GB, IQ3_XS is 25.7 GB, Q3_K_S is 27.1 GB, IQ3_S is 27.1 GB, IQ3_M is 28.0 GB, Q3_K_M is 30.1 GB and 6 more.
What context length does Llama3.1 60B support?
The GGUF header lists a trained context length of 8,192 tokens (8K). The KV cache grows with the context you use: about 2.3 GB at 8K and 2.3 GB at 8K.
Where can I download Llama3.1 60B GGUF?
The files are on Hugging Face in the mradermacher/Llama3.1-60B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Llama3.1 60B use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.