Llama 3.2 3B
By Meta · United States
Updated 2026-07-13
Overview
Meta's 3B instruct model with a full 128k context, tuned for laptops, mobile, and edge devices where memory and battery matter.
When to pick this model
- On-device assistants for laptops, phones, or tablets
- CPU-only inference where speed beats raw quality
- Long-context summarization on constrained hardware
- Latency-critical agent loops
- Local autocomplete or text classification
VRAM requirements by quantization
| Quantization | VRAM required |
|---|---|
| Q4_K_M (recommended) | 2.5 GB |
| Q5_K_M | 3 GB |
| Q8_0 | 4.5 GB |
| FP16 (no quantization) | 7 GB |
VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.
In practice, Llama 3.2 3B fits an 8 GB consumer card at Q4_K_M (2.5 GB). Stepping up to Q8_0 nearly doubles the footprint to 4.5 GB, and unquantized FP16 weights take 7 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.
Without a GPU, Llama 3.2 3B needs roughly 6 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 25 tokens/sec on entry-level GPUs, on the order of 70 tokens/sec on a mid-range card, and up to 160 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.
What hardware do you need
The table below matches Llama 3.2 3B to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.
| GPU memory | Example cards | Best fit for Llama 3.2 3B |
|---|---|---|
| 8 GB | RTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GB | FP16 (7 GB used) |
| 12 GB | RTX 5070, RTX 5070 Ti Laptop, RTX 4080 Laptop | FP16 (7 GB used) |
| 16 GB | RTX 5080, RTX 4080 Super, Radeon RX 9070 XT | FP16 (7 GB used) |
| 24 GB | RTX 4090, Radeon RX 7900 XTX, RTX 5090 Laptop | FP16 (7 GB used) |
| 32 GB | RTX 5090 | FP16 (7 GB used) |
Which GPU should you buy to run Llama 3.2 3B?
To run Llama 3.2 3B locally at Q4, you need ~2.5 GB of VRAM. The best value for this is a RTX 5060 (8 GB VRAM).
As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.
Published benchmark scores
| Benchmark | Score |
|---|---|
| MMLU | 63.4 |
| HellaSwag | 79.2 |
Scores published by the model author or aggregated from public leaderboards. Re-measured monthly by our editorial team.
To put Llama 3.2 3B in context: its MMLU score of 63.4 ranks #29 of the 34 catalog models with a published MMLU result (catalog median 73.4). Scores are author-published and measured under different harnesses, so treat gaps of a few points as noise rather than a verdict.
Strengths
- 128k context in a 3B parameter footprint
- Fast CPU inference
- Strong baseline for edge and mobile use cases
- Distilled from larger Llama models for better quality density
Limitations
- Noticeably weaker than 7B+ models on complex tasks
- No vision in this checkpoint
- Subject to Llama Community license terms
Typical workloads
In our catalog grid, Llama 3.2 3B is filed under On-device, Fast Assistants — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: everyday chat, drafting and summarization.
The 128k-token context window is large enough to hold entire codebases' worth of files or long reports in a single prompt, which is what makes local RAG and document analysis practical. It ships under the Llama 3 Community license — commercial use is generally possible but read the specific terms before embedding it in a product.
Architecture & training
Architecture: Dense Transformer · Llama 3.2 3B · lightweight architecture for edge
Training: Meta multilingual corpus + distillation from larger Llama models.
The best 3B open-weight model for edge use cases — pick it when memory and latency dominate the brief.
Quick start
ollama run llama3.2:3bOr use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.
Similar models worth comparing
Frequently asked questions
How much VRAM does Llama 3.2 3B need?
At the recommended Q4_K_M quantization, Llama 3.2 3B needs about 2.5 GB of VRAM. Q8_0 takes 4.5 GB, and unquantized FP16 weights take 7 GB.
Can Llama 3.2 3B run without a GPU?
Yes — with roughly 6 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.
What context window does Llama 3.2 3B support?
Llama 3.2 3B supports a 128k-token context window (131,072 tokens).
Can I use Llama 3.2 3B commercially?
Llama 3.2 3B ships under the Llama 3 Community license. Commercial use is generally permitted subject to its terms — review the license text before shipping a product.
How fast is Llama 3.2 3B on consumer hardware?
Our compatibility engine estimates on the order of 70 tokens/sec on a mid-range GPU and up to 160 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.
Which quantization of Llama 3.2 3B should I download first?
Start with Q4_K_M (2.5 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. It already fits an 8 GB card at FP16.