From the model card
Fields below are copied from the tags and counters on the HuggingFace repository janhq/Jan-v3.5-4B-gguf at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- janhq
- Pipeline tag
- text-generation
- Library
- Transformers
- Weight formats
- GGUF
- License tag
apache-2.0— read the license file in the repo before relying on it- Lineage
-
- base model janhq/Jan-v3-4B-base-instruct
- quantized from janhq/Jan-v3-4B-base-instruct
- Language tags
- English (en)
- Downloads (HF counter at last fetch)
- 336,972
- Likes (HF counter at last fetch)
- 21
- Model card
- https://huggingface.co/janhq/Jan-v3.5-4B-gguf
Use cases
- Local private AI assistant on a consumer laptop via the Jan desktop app
- On-device math and coding help without cloud API dependency
- Building llama.cpp-based local assistants with a well-tuned small model
- Privacy-sensitive professional tasks where cloud data upload is unacceptable
- Comparing local 4B instruction models against cloud API alternatives
Pros
- GGUF format; straightforward loading in Jan app and llama.cpp
- Jan.ai maintained; regular updates and desktop app integration
- 4B size runs on most laptops with 8GB+ RAM
- Math and coding fine-tuning improve on base model task performance
Cons
- Optimised for Jan.ai's specific interface; behaviour in other runtimes may differ
- 4B ceiling limits reasoning depth for complex professional tasks
- No explicit license beyond Jan.ai terms; verify redistribution rights
- 16 likes; limited external benchmarking vs other 4B GGUF alternatives
Tags
transformersggufmathidentitytext-generationenbase_model:janhq/Jan-v3-4B-base-instructbase_model:quantized:janhq/Jan-v3-4B-base-instructlicense:apache-2.0endpoints_compatibleregion:usconversational