# tinygrad

> neural network framework

**Wikidata**: [Q117105889](https://www.wikidata.org/wiki/Q117105889)  
**Source**: https://4ort.xyz/entity/tinygrad

## Summary
tinygrad is a neural network framework classified as a type of software. It was created by George Hotz, an American hacker and programmer, and is developed by the company tiny corp. The framework is open-source, distributed under the MIT License, and is accessible via its website and GitHub repository.

## Key Facts
- **Entity Type:** Software / Neural Network Framework.
- **Creator:** George Hotz (American hacker, programmer, and computer scientist).
- **Developer:** tiny corp.
- **License:** MIT License.
- **Website:** https://tinygrad.org/
- **Source Code:** https://github.com/geohot/tinygrad (later https://github.com/tinygrad/tinygrad).
- **Latest Stable Version:** 0.10.1 (released February 5, 2025).
- **Development Status:** Active (frequent stable releases recorded between 2022 and 2025).
- **Copyright Status:** Copyrighted.

## FAQs
### Q: Who created tinygrad?
A: tinygrad was created by George Hotz, an American hacker, programmer, and computer scientist also known as "geohot."

### Q: What license is tinygrad released under?
A: tinygrad is released under the MIT License, a permissive free software license originating from the Massachusetts Institute of Technology.

### Q: When was the most recent version of tinygrad released?
A: According to the provided structured data, the most recent stable version is 0.10.1, which was released on February 5, 2025.

## Why It Matters
tinygrad serves as a significant tool in the domain of artificial intelligence and machine learning as a distinct neural network framework. Its relevance is underscored by its development lineage, created by George Hotz, a notable figure in the hacking and programming communities. The project has evolved from a personal repository (`geohot/tinygrad`) to an organization-managed project (`tinygrad/tinygrad`) under the entity tiny corp, indicating professional commercial development and support.

The framework is maintained with high frequency, evidenced by a rapid release schedule that transitioned from version 0.4.0 in late 2022 to version 0.10.1 by early 2025. This consistent iteration suggests an active effort to refine neural network capabilities and maintain compatibility with modern hardware and software environments. By offering an MIT License, it provides developers and researchers with permissive rights to use, modify, and distribute the software, facilitating broad adoption and integration into various applications.

## Notable For
- **Creator Pedigree:** Developed by George Hotz, a recognized American hacker and computer scientist.
- **Rapid Release Cycle:** Maintained a consistent schedule of stable releases (versions 0.7.0 through 0.10.1) roughly every few months from 2023 to 2025.
- **Open Source Accessibility:** Utilizes the permissive MIT License, allowing for wide usage and modification.
- **Organizational Backing:** Supported and developed by a dedicated corporate entity, tiny corp.

## Body

### Development and Origins
tinygrad is a neural network framework authored by George Hotz, an American programmer and computer scientist born on October 2, 1989. The project is classified as executable software and is actively developed by the organization **tiny corp**.

The source code was originally hosted at `https://github.com/geohot/tinygrad` but is referenced under the organization `tinygrad` in later release data.

### Licensing and Availability
The software is distributed under the **MIT License**. It is available to the public via its official website `https://tinygrad.org/` (English language) and its GitHub repositories. The project retains a copyrighted status.

### Release History
The framework has undergone significant versioning updates, moving from early iterations to stable releases with increasing regularity.

**Early Versions:**
- **0.4.0:** Released November 8, 2022.
- **0.5.0:** Released March 7, 2023.
- **0.6.0:** Released May 26, 2023.

**Stable Releases (2023–2024):**
- **0.7.0:** Marked as a stable version, released August 27, 2023.
- **0.8.0:** Released January 9, 2024.
- **0.9.0:** Released May 28, 2024.
- **0.9.1:** Released June 29, 2024.
- **0.9.2:** Released August 13, 2024.
- **0.10.0:** Released November 19, 2024.

**Recent Versions (2025):**
- **0.10.1:** Designated as a stable version, released on February 5, 2025.

## References

1. [Source](https://api.github.com/repos/geohot/tinygrad)
2. [Release 0.4.0. 2022](https://github.com/geohot/tinygrad/releases/tag/v0.4.0)
3. [Release 0.5.0. 2023](https://github.com/geohot/tinygrad/releases/tag/v0.5.0)
4. [Release 0.6.0. 2023](https://github.com/geohot/tinygrad/releases/tag/v0.6.0)
5. [Release 0.7.0. 2023](https://github.com/tinygrad/tinygrad/releases/tag/v0.7.0)
6. [Release 0.8.0. 2024](https://github.com/tinygrad/tinygrad/releases/tag/v0.8.0)
7. [Release 0.9.0. 2024](https://github.com/tinygrad/tinygrad/releases/tag/v0.9.0)
8. [Release 0.9.1. 2024](https://github.com/tinygrad/tinygrad/releases/tag/v0.9.1)
9. [Release 0.9.2. 2024](https://github.com/tinygrad/tinygrad/releases/tag/v0.9.2)
10. [Release 0.10.0. 2024](https://github.com/tinygrad/tinygrad/releases/tag/v0.10.0)
11. [Release 0.10.1. 2025](https://github.com/tinygrad/tinygrad/releases/tag/v0.10.1)
12. [Release 0.10.2. 2025](https://github.com/tinygrad/tinygrad/releases/tag/v0.10.2)
13. [Release 0.10.3. 2025](https://github.com/tinygrad/tinygrad/releases/tag/v0.10.3)
14. [Release 0.11.0. 2025](https://github.com/tinygrad/tinygrad/releases/tag/v0.11.0)
15. [tinygrad 0.12.0. 2026](https://github.com/tinygrad/tinygrad/releases/tag/v0.12.0)