Model catalog

Llama 3.2 3B Instruct

Meta's 3B instruction-tuned Llama model, suitable for modest local hardware when quantized.

Check this model on your hardware See what your PC can run

Model overview

Family: Llama 3.2
Provider: Meta
Architecture: Llama
Parameters: 3B
Formats: gguf, safetensors
Runtimes: llama.cpp
License: Llama 3.2 Community License
Default context: 4,096 tokens
Maximum context: 131,072 tokens

Available quantizations

Lower-bit quantizations generally use less memory. The compatibility calculator uses these candidates and its transparent approximate memory assumptions; it does not make a performance guarantee.

Q4_K_M

Common lower-memory GGUF choice; actual files are produced by community or vendor conversion pipelines.

Bits per weight: 4.5
Source: llama.cpp quantization documentation

Q5_K_M

Middle-ground GGUF choice with a larger memory requirement than Q4_K_M.

Bits per weight: 5.5
Source: llama.cpp quantization documentation

Q8_0

Higher-memory GGUF choice; it is still an approximate memory-planning input rather than a performance claim.

Bits per weight: 8
Source: llama.cpp quantization documentation

Compatibility notes

Weight estimates are approximate and include a runtime overhead. Real requirements vary with context length, runtime, drivers, and other system use. A model page cannot determine compatibility without your hardware profile; use the calculator for that assessment.

Meta Llama 3.2 model card; verified value, last verified 2026-09-22. Publisher metadata is separated from approximate GGUF memory-planning candidates. View source