---
license: other
license_name: commercial
license_link: https://hanzo.ai/contact
language:
- en
tags:
- agentic
- coding
- llm
- training
- claude
- programming
size_categories:
- 1B<n<10B
---

# Zen Agentic Dataset

**8.47 Billion Tokens** of real-world agentic AI programming, blockchain development, and cutting-edge infrastructure code.

## Dataset Overview

A comprehensive training dataset combining Claude Code interactions with full git history from 1,400+ repositories spanning 15 years of professional development.

| Metric | Value |
|--------|-------|
| **Total Tokens** | 8.47 billion |
| **Training Samples** | 3.35 million |
| **Validation Samples** | 100,000 |
| **Total Size** | ~27 GB |
| **Repositories** | 1,452 |
| **Time Span** | 15 years (2010-2025) |

## Data Composition

| Component | Tokens | Percentage |
|-----------|--------|------------|
| Claude Code Debug Sessions | 2.42B | 29% |
| Claude Conversations | 1.14B | 13% |
| Claude Interactions | 0.86B | 10% |
| Git History | 4.03B | 48% |

## Domain Coverage

### Agentic AI & LLM Infrastructure
- Model Context Protocol (MCP) - 260+ tool implementations
- Multi-agent orchestration - Claude, GPT-4, Gemini integrations
- Agent frameworks - Planning, memory, tool use, reflection

### Web3 & Blockchain
- Smart contracts - Solidity, Vyper (ERC20, ERC721, DeFi)
- Consensus engines - Snow family, BFT, DAG-based protocols
- Cross-chain bridges and DeFi protocols

### Cryptography & Security
- Post-quantum cryptography implementations
- Threshold cryptography and MPC
- Zero-knowledge proofs experimentation

### Modern Development
- Full-stack TypeScript - Next.js 14+, React 18+
- Systems programming - Rust, Go, Python, C/C++
- DevOps - Docker, Kubernetes, CI/CD

## Licensing & Access

**This dataset is available for research and commercial licensing.**

### For Developers & Researchers

We award grants to individuals and teams who want to train models on this dataset, particularly those building:
- Models for specific blockchain ecosystems
- Open-source AI tools using OpenAI-compatible protocols
- Research advancing agentic AI capabilities

### To Request Access

**Contact:** z@hanzo.ai

Please include:
- Intended use case (training, research, evaluation)
- Organization/affiliation
- Target ecosystem (if applicable)
- Licensing requirements

### Supported Organizations

Dataset mirrors are maintained by:
- [Hanzo AI](https://hanzo.ai) - AI infrastructure platform
- [Lux Network](https://lux.network) - AI compute settlement layer
- [Zen LM](https://zenlm.org) - Open model research
- [Zoo Labs](https://zoo.ngo) - Decentralized AI research

## Models Trained on This Dataset

| Model | Size | Architecture | Status |
|-------|------|--------------|--------|
| Zen Coder 4B | 4B | Qwen3 | Trained |
| Zen Coder 24B | 24B | Devstral Small 2 | Trained |
| Zen Coder 123B | 123B | Devstral 2 | Training |
| Zen Coder Max | 358B | GLM-4.7 (MoE) | Planned |
| Zen Coder Ultra | 1T | Kimi K2 (MoE) | Planned |

## Training Framework

Use [Zen Trainer](https://github.com/zenlm/zen-trainer) for fine-tuning:

```python
from zen_trainer import ZenTrainer

trainer = ZenTrainer(
    model_key="qwen3-4b",
    dataset_path="hanzoai/zen-agentic-dataset-private",  # Requires access
    output_dir="./output/my-model",
)
trainer.train()
```

## Related Projects

- [Zen Trainer](https://github.com/zenlm/zen-trainer) - Training framework
- [Hanzo MCP](https://github.com/hanzoai/mcp) - Model Context Protocol (260+ tools)
- [Hanzo AI](https://hanzo.ai) - AI infrastructure platform
- [Lux Network](https://lux.network) - AI compute settlement layer
- [Zoo Labs](https://zoo.ngo) - Decentralized AI research

## Citation

```bibtex
@dataset{zen_agentic_dataset,
  author = {Kelling, Zach},
  title = {Zen Agentic Dataset: 8.47B Tokens of Agentic AI Programming},
  year = {2025},
  publisher = {Zoo Labs Foundation},
  url = {https://huggingface.co/datasets/hanzoai/zen-agentic-dataset}
}
```

---

**Maintainer:** z@hanzo.ai
**License:** Commercial - Contact for licensing terms