Add PaperClaw: AI-Powered Scientific Paper Generation VS Code Extension

Proposal: Add CAJAL — Local Scientific Paper Generation LLM

CAJAL is a fine-tuned open-source LLM (Qwen3.6 base) specialized for generating publication-ready scientific papers, running 100% locally via Ollama/llama.cpp.

What it does

• Generates complete 7-section papers (Abstract, Introduction, Methodology, Results, Discussion, Conclusion, References)
• All citations verified against real arXiv papers via API
• Optional AI Tribunal scoring (3 reviewers, 0-10 scale, iterative rewrite)
• Runs offline on consumer GPUs (RTX 3090, Mac M-series)

Sizes available

Model	Size	Format	Use Case
CAJAL-4B	~2.3GB	Q4_K_M	Fast drafting, laptops
CAJAL-9B	~4.5GB	Q4_K_M	Full papers, desktops
CAJAL-9B	~5.5GB	Q5_K_M	Maximum quality

Links

• GitHub: https://github.com/Agnuxo1/CAJAL
• HuggingFace: https://huggingface.co/Agnuxo/CAJAL-4B-P2PCLAW
• Paper: https://arxiv.org/pdf/2604.19792
• Ollama Modelfile: https://github.com/Agnuxo1/CAJAL/blob/main/Modelfile

Why add it

CAJAL fills a gap in the local LLM ecosystem: no other open-source model is specifically fine-tuned for academic paper generation with real citation verification. It enables researchers to draft papers privately without sending data to cloud APIs.

Happy to adapt this entry to your list format!

— Francisco (@Agnuxo1), P2PCLAW