Jan v3 4B: Compact Model Excels at Math & Code
Jan v3 4B is a compact language model that demonstrates strong performance in mathematical reasoning and code generation tasks despite its smaller parameter
Someone from the Jan team released a 4B-parameter model that punches above its weight for math and coding tasks.
To try it:
Download Jan Desktop from https://www.jan.ai/, then grab the model from Jan Hub, or pull it directly:
- Jan-v3-4B: https://huggingface.co/janhq/Jan-v3-4B-base-instruct
- GGUF version: https://huggingface.co/janhq/Jan-v3-4B-base-instruct-gguf
They recommend these settings for best results:
temperature: 0.7
top_p: 0.8
top_k: 20Pretty solid for a lightweight model - runs locally without needing a beefy GPU. Works well as a base for fine-tuning if someone wants to customize it for specific tasks. The team’s planning bigger releases too, including a 30B version and specialized code/search variants.
Good option for anyone wanting local AI that doesn’t destroy their system resources.
Related Tips
KaniTTS2: Fast Local Text-to-Speech with Cloning
KaniTTS2 provides a fast, locally-run text-to-speech system with voice cloning capabilities, enabling users to generate natural-sounding speech from text while
AdaLLM: True FP4 Inference on RTX 4090s Without FP16 Fallbac
AdaLLM enables genuine 4-bit floating-point inference on RTX 4090 GPUs without reverting to 16-bit precision, delivering faster and more memory-efficient large
Chatbot Framework Rebuilt in Rust: 10MB Binary
A chatbot framework originally written in another language has been completely rewritten in Rust, resulting in a remarkably compact 10MB binary that