作者: Jakub Rusinowski · 最后更新: 2026年8月15日
Meta的多模态和边缘优化系列。Llama 3.2引入视觉能力(11B和90B变体),以及专为智能手机和边缘设备设计的超紧凑1B/3B版本。所有变体均支持128k上下文窗口。
| Licence | What it permits | Applies to |
|---|---|---|
Llama Community | Commercial use permitted Weights are downloadable and commercial use is permitted, subject to the licence’s acceptable-use terms. | Llama 3.2 1B Instruct, Llama 3.2 3B Instruct, Llama 3.2 11B Vision Instruct, Llama 3.2 90B Vision Instruct |
| Llama 3.2 1B Instruct | Min 2 GB VRAM · Q4_K_M · 128,000 ctx · ollama run llama3.2:1b |
| Llama 3.2 3B Instruct | Min 3 GB VRAM · Q4_K_M · 128,000 ctx · ollama run llama3.2:3b |
| Llama 3.2 11B Vision Instruct | Min 7 GB VRAM · Q4_K_M · 128,000 ctx · |
| Llama 3.2 90B Vision Instruct | Min 54 GB VRAM · Q4_K_M · 128,000 ctx · |
The cheapest GPU that runs Llama 3.2 Family locally (min 2 GB VRAM) is the Intel Arc B570 (10 GB).
Install Ollama then run: ollama run llama3.2:1b
Minimum VRAM: 2 GB. For best results use Q4_K_M quantization.
Llama 3.2 Family needs about 2 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: Llama 3.2 1B Instruct (2 GB, Q4_K_M); Llama 3.2 3B Instruct (3 GB, Q4_K_M); Llama 3.2 11B Vision Instruct (7 GB, Q4_K_M); Llama 3.2 90B Vision Instruct (54 GB, Q4_K_M). On Apple Silicon, unified memory counts toward this requirement.
Yes — Llama 3.2 Family runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
Q4_K_M is the best balance of quality and VRAM for Llama 3.2 Family in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
Install Ollama, then run: ollama run llama3.2:1b. This downloads Llama 3.2 Family and starts a local, OpenAI-compatible endpoint — no internet connection is needed after the initial download.