Meta · open-weights model
Llama-3.1-70B-Instruct
Llama-3.1-70B-Instruct is a 70.6B-parameter model by Meta. Its smallest quant (Q4_K_M) needs at least 45.8GB of GPU-addressable memory at 8K context; it loads cleanly at a quality quant on 7 of the 50 machines we track.
- Parameters
- 70.6B
- Layers
- 80
- Max context
- 128K
- Attention
- GQA
- License
- llama3.1
Config from meta-llama/Llama-3.1-70B-Instruct
Quants and what they need
Measured file sizes; “needs” = weights + KV cache at 8K context + overhead.
| Quant | File size | Needs ≥ |
|---|---|---|
| Q8_0 | 75.0 GB | 78.3 GB |
| Q6_K | 57.9 GB | 61.2 GB |
| Q5_K_M | 50.0 GB | 53.3 GB |
| Q4_K_M | 42.5 GB | 45.8 GB |
What runs Llama-3.1-70B-Instruct
Every machine we track, graded at its best clean-fit quant.
Verdicts computed from published specs and measured GGUF sizes — how the math works.