Llama 3.2 (3B)
Meta's 3B chat model. Small enough to run on modest hardware, sharp enough to be a useful default.
At a glance
- Parameters
- 3B
- Quantization
- Q4_K_M
- Context window
- 128K tokens
- Approx. size
- 2.0 GB
- Engines
- llama.cpp, MLX, Ollama
- License
- Llama 3.2 Community License (custom-restricted)
- Version
- v1.1.0 , released 2026-04-26
- Family
- llama
Tuning
Defaults a gezel client applies out of the box, from real evaluation runs against this exact quantization.
- sampling.temperature
0.6- sampling.topP
0.9- profiles
- thinking-general, thinking-coding, thinking-precise, instruct, creative
Model behaviors
Client-side behavior modules the catalog enables for this model (reasoning-tag handling, fabrication detection, prompt shaping).
- fabrication.detect-past-tense-no-tools
Sources
- llama.cpp (GGUF)
- bartowski/Llama-3.2-3B-Instruct-GGUF ·
Llama-3.2-3B-Instruct-Q4_K_M.gguf· Q4_K_M - MLX
- mlx-community/Llama-3.2-3B-Instruct-4bit
- Ollama
llama3.2- Upstream
- https://ollama.com/library/llama3.2
- Manifest
- View on GitHub
Tags
- meta
- small