Llama 3.2 family: which size to run
2 sizes, 1.23B–3.21B parameters · Meta
Sizes
| Model | Architecture | |||||||
|---|---|---|---|---|---|---|---|---|
| Llama 3.2 1B Instruct | 1.23B | Dense | 128,000 | No LiveBench result for this model in release 2026-06-25 | 9 | 3 | 0.8 | 0.7 |
| Llama 3.2 3B Instruct | 3.21B | Dense | 128,000 | No LiveBench result for this model in release 2026-06-25 | 11 | 5 | 2 | 1.8 |
Each size is its own model, with its own scores and memory needs. Q4 estimates are planning figures, not measurements; file sizes are published. The Mac or PC column is for a 32 GB or larger machine (8 GB reserve); machines under 32 GB reserve 25% of their memory (2 to 8 GB), so they need less. The publisher lists 1B, 3B (text in, text out).
How to read this table
Q4 means compressed to about 4 bits per weight. GGUF (llama.cpp, Ollama, LM Studio) and MLX (Apple silicon) are file formats; the sizes shown are the published 4-bit files. A dash means no value is published or collected for that size; the reason is in the cell's accessible text.
Which size fits my machine
The largest size whose Q4 planning estimate fits what each machine offers. A planning estimate, not a measurement and not a speed.
A dash means no size of this family fits that machine. The requirement is the size's Q4 estimate with that machine's own reserve: a graphics card reserves 2 GB, as in the sizes table; a Mac or PC reserves 25% of its memory (2 to 8 GB), so below 32 GB its figure is lower than the sizes table's. A server profile reserves 2 GB per graphics card, so its figure is higher.
Smallest machine for each size
The profile with the least memory that still has enough for the size's Q4 estimate, from the hardware catalog.
| Size | Smallest Mac or PC-memory profile | Smallest graphics-card profile |
|---|---|---|
| Llama 3.2 1B Instruct | Mac mini 2020 · M1 · 8 GB | GTX 1650 · 4 GB |
| Llama 3.2 3B Instruct | Mac mini 2020 · M1 · 8 GB | RTX 3050 · 6 GB |
A dash means no profile of that kind in the catalog has enough memory for the size.
Memory by file precision
The same planning formula at three precisions (0.625, 1.125 and 2 bytes per parameter, plus working room and the system reserve). Estimates, not measurements.
| Size | ||||||
|---|---|---|---|---|---|---|
| Llama 3.2 1B Instruct | 9 | 3 | 10 | 4 | 11 | 5 |
| Llama 3.2 3B Instruct | 11 | 5 | 13 | 7 | 16 | 10 |
Family tree
Family, sizes and, for each size, the two 4-bit files the tables use (GGUF Q4_K_M and MLX 4-bit) when one was collected. Other precisions are not drawn here.
The same tree as a list:
- Llama 3.2 · Meta
- 1B Instruct
- GGUF Q4_K_M · 0.8 GB · bartowski/Llama-3.2-1B-Instruct-GGUF, read Oct 23, 2026
- MLX 4-bit · 0.7 GB · mlx-community/Llama-3.2-1B-Instruct-4bit, read Oct 23, 2026
- 3B Instruct
- GGUF Q4_K_M · 2 GB · bartowski/Llama-3.2-3B-Instruct-GGUF, read Oct 23, 2026
- MLX 4-bit · 1.8 GB · mlx-community/Llama-3.2-3B-Instruct-4bit, read Oct 23, 2026
- 1B Instruct
Learn more
See Llama 3.2 1B Instruct
The smallest size of the family: its scores, memory estimate and the machines it runs on.
Sources
Where the family and its sizes are named, and where each number comes from.
- Meta Llama 3.2 model card · “The Llama 3.2 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction-tuned generative models in 1B and 3B sizes (text in/text out).” · read Oct 17, 2026
- Meta Llama 3.2 model card · “Llama 3.2 1B and 3B models are smaller and less capable derivatives of Llama 3.1.” · read Oct 17, 2026
- Llama 3.2 1B Instruct model card
- Llama 3.2 3B Instruct model card
- LiveBench 2026-06-25 · the only quality source on this page; a dash means the release does not include the model · snapshot collected Sep 26, 2026
- Published file sizes: Hugging Face repositories, read Oct 23, 2026
- Methodology · how the Q4 planning estimate is computed · Learn