Llama 3.2 3B

Name: Llama 3.2 3B
Price: 0.12 USD
Author: Meta

Meta · Llama 3 · released 2024-09-25 · data checked 2026-11-01 Needs verification

Llama 3.2 3B is a 3.2B-parameter dense model from Meta (United States), released 2024-09-25 under the Llama 3.2 Community license. It doesn't have enough reported benchmarks for an index score yet. It needs about 3.4 GB of memory at 4-bit with an 8K-token context and costs $0.12 per million tokens through the cheapest API provider.

Small LM Index

—

Needs 2 core benchmarks (has 1)

Parameters

3.2B

Dense

Price per 1M tokens

$0.12

$0.050 in / $0.33 out · blended 3:1

Context window

128K

tokens

Benchmarks

Each bar shows this model's score. The tick marks the best tiny model on that benchmark. Sources: 1 reported by Meta. How we label sources.

GPQA Diamond Lab-reported Knowledge32.8

Class best: Nanbeige4.2 3B 87.4

Specifications

Lab: Meta (US · United States)
Architecture: Dense transformer
Total parameters: 3.2B
Active parameters: All (dense)
Size class: Tiny
Context window: 128K tokens
Input modalities: text
Reasoning: Instruct (no thinking)
License: Llama 3.2 Community
Openness: Open weights
Memory to run: ~3.4 GB at 4-bit, 8K context (more sizes)
Hugging Face: 969.6K downloads (30 days) · 2.9K likes
API providers: 2 on OpenRouter
API price: $0.050 in / $0.33 out per 1M (OpenRouter, 2026-11-03)

Weights on Hugging Face ↗Try on OpenRouter ↗

Compare with

Llama 3.2 3B vs SmolLM3 3B (3.1B, index 30.4)
Llama 3.2 3B vs Granite 4.2 3B (3B, index 54.8)
Llama 3.2 3B vs Ministral 3 3B (3B, index 49.4)

Sources

https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct

Something wrong here? Report it.

Run it locally

Uses unsloth/Llama-3.2-3B-Instruct-GGUF (Q4_K_M). Swap the tag for another size: BF16, F16, IQ4_NL, IQ4_XS, Q2_K, Q2_K_L, Q3_K_M, Q3_K_S….

Ollama

ollama run hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M

llama.cpp

llama-server -hf unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M

Then open http://localhost:8080 for a chat window.

vLLM (GPU server)

vllm serve meta-llama/Llama-3.2-3B-Instruct

Full-precision weights, for a server GPU.

Open in LM Studio

Memory needed: weights, context cache and runtime buffers.

Quantization	4K context	32K context	128K context
16-bit F16	7.4 GB	11 GB	22 GB
8-bit Q8_0	4.4 GB	7.7 GB	19 GB
4-bit Q4_K_M	3.0 GB	6.3 GB	17 GB

Check it against your GPU, Mac or phone →