Jan-Nano: An Agentic Model

Note: Jan-Nano is a non-thinking model.


Reckoning Quantization

This model directory was generated by Reckoning Q.O.D. (Quantize On Demand) from a local source model.

Reckoning rewrote selected matrix weights into its Q4 quantized tensor format to reduce model size and improve local inference throughput while keeping the model card and tokenizer metadata intact.

What Changed

  • Original local size: 7.5 GB
  • Quantized local size: 2.9 GB
  • Target dtype: Q4
  • Quantized tensors: 252
  • Provenance manifest: reckoning-quantization.json

The manifest records the source/output paths, tensor dtype changes, generated .qb sidecars, and shape-normalization transforms used during export.

Try Reckoning

Reckoning is a local Java inference engine with safetensors loading, Q4 quantization, and quantize-on-demand model generation. Learn more at github.com/edwardcapriolo/reckoning.

GitHub

Jan-Nano

Authors: Alan Dao, Bach Vu Dinh

Overview

Jan-Nano is a compact 4-billion parameter language model specifically designed and trained for deep research tasks. This model has been optimized to work seamlessly with Model Context Protocol (MCP) servers, enabling efficient integration with various research tools and data sources.

Evaluation

Jan-Nano has been evaluated on the SimpleQA benchmark using our MCP-based benchmark methodology, demonstrating strong performance for its model size:

image/png

The evaluation was conducted using our MCP-based benchmark approach, which assesses the model's performance on SimpleQA tasks while leveraging its native MCP server integration capabilities. This methodology better reflects Jan-Nano's real-world performance as a tool-augmented research model, validating both its factual accuracy and its effectiveness in MCP-enabled environments.

How to Run Locally

Jan-Nano Demo

Jan-Nano is currently supported by Jan, an open-source ChatGPT alternative that runs entirely on your computer. Jan provides a user-friendly interface for running local AI models with full privacy and control.

For non-jan app or tutorials there are guidance inside community section, please check those out! Discussion

VLLM

Here is an example command you can use to run vllm with Jan-nano

vllm serve Menlo/Jan-nano --host 0.0.0.0 --port 1234 --enable-auto-tool-choice --tool-call-parser hermes --chat-template ./qwen3_nonthinking.jinja

Chat-template is already included in tokenizer so chat-template is optional, but in case it has issue you can download the template here Non-think chat template

Recommended Sampling Parameters

  • Temperature: 0.7
  • Top-p: 0.8
  • Top-k: 20
  • Min-p: 0

๐Ÿ“„ Citation

@misc{dao2025jannanotechnicalreport,
      title={Jan-nano Technical Report}, 
      author={Alan Dao and Dinh Bach Vu},
      year={2025},
      eprint={2506.22760},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2506.22760}, 
}

Documentation

Setup, Usage & FAQ

Downloads last month
249
Safetensors
Model size
0 params
Tensor type
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for edwardcapriolo/Jan-nano

Finetuned
Qwen/Qwen3-4B
Finetuned
(1,300)
this model

Paper for edwardcapriolo/Jan-nano