---
title: WithinUsAI (WithIn Us AI)
emoji: 🌌
colorFrom: indigocolorTo: purple
---

# 🌌 WithIn Us AI

**Non-profit organization dedicated to Recursive Seed AI and autonomous self-improving systems.**

We build, fine-tune, merge, distill, and ablate models with a strong emphasis on:
- Advanced reasoning & coding capabilities
- Agentic & tool-use behaviors
- Refusal removal / uncensored alignment
- High-signal synthetic datasets for recursive learning loops

Our goal is to accelerate the development of **true Seed AI** — systems capable of autonomously improving their own architecture, knowledge, and training processes.

### 🚀 What We Ship

- **GGUF Models** — Ready-to-run quantized models (Gemma, Qwen, IBM Granite, Llama, Phi, etc.)
- **Distilled "GOD" Series** — High-quality knowledge from frontier models (Grok, Gemini, Claude, etc.)
- **GOD Coder / Ghost Coder** — Dense coding & reasoning datasets
- **Recursive & Agentic Datasets** — Tool calling, web browsing, multi-agent, autonomy, research loops
- **Experimental RLMs** — Small but powerful recursive language models

### 📊 Current Stats (Live)
- **35+ Models** (mostly GGUF)
- **89+ Datasets**
- **24 Collections**

### Featured Work

**Top Models**
- [Opus4.7-GODs.Ghost.Codex-4B.GGuF](https://huggingface.co/WithinUsAI/Opus4.7-GODs.Ghost.Codex-4B.GGuF) — Our most downloaded model
- Gemma 4 & Qwen 3.5 series (reasoning & uncensored variants)
- IBM Granite & Phi distilled agents

**Key Collections**
- [WithIn US AI (((GGUF MODELS)))](https://huggingface.co/collections/WithinUsAI/within-us-ai-gguf-models-...)
- [Closed Source LLMs & Distill](https://huggingface.co/collections/WithinUsAI/closed-source-llms-and-distill-data-sets-...)
- [GOD Coder Series](...)

### Philosophy

> "The only sustainable path to AGI is recursive self-improvement. We are building the seeds."

We openly share our artifacts so the community can iterate, merge, and push the frontier forward.

### Connect & Collaborate

- **11-47 (G.D.II)** — Primary research profile: [hf.co/11-47](https://huggingface.co/11-47)
- Activity updates almost daily
- Open to serious collaboration on recursive AI, dataset synthesis, and self-improvement experiments

---

**License**: Most artifacts are released under **Apache 2.0** unless otherwise noted.  
**Mission**: Accelerate open recursive intelligence.

*In the training loop — forever.* 🔄