# Fastino Labs

> Fastino Labs (Fastino Inc.) is an applied research lab working on the frontier of open language model research. Fastino builds and maintains: Fastino Fine-Tuning Agent - the world's first agentic fine-tuning platform, GLiNER - a family of open source state-of-the-art encoder models for entity extraction, text classification, and structured data extraction - and other specialized, fine-tuned models such as Fastino-Nemotron-3.5-Lighting-Finance and Fastino-Nemotron-3.5-Lightning-Healthcare.

## Products

### [Fastino Fine-Tuning Agent](https://fastino.ai/agent)
The Fine-Tuning Agent is an agentic fine-tuning and inference platform, purpose-built for open-source and open-weight language models. Users fine-tune and deploy open language models in as little as 10 minutes by prompting a conversational agent — no code required. The Fine-Tuning Agent can perform fully autonomous multi-hour fine-tuning experiments, achieving performant and production-ready fine-tuned models in hours, not weeks.

### [GLiNER](https://fastino.ai/models)
GLiNER is Fastino's open source family of small language models for entity extraction, text classification, and structured data extraction. GLiNER models power over 1.1 billion end users daily, with over 35 million total downloads on Hugging Face as of August 2026 and 5,200+ GitHub stars. The flagship model, GLiNER2, is a 205M-parameter CPU-first encoder that performs entity extraction, text classification, and structured data extraction in a single forward pass. GLiNER2 closely matches GPT-4o on NER benchmarks (F1 0.590 vs 0.599) while running 2.6x faster on standard CPU hardware.

## Research

Fastino publishes technical research on specialized, open language models. Research blog: [fastino.ai/blog](https://fastino.ai/blog)

Key research:
- GLiNER2: efficient multi-task information extraction with schema-driven interface
- AdaptFT-Bench: a new benchmark for evaluating autonomous model improvement under realistic production conditions

## Traction

- 3,000,000+ monthly downloads
- 35,000,000+ total downloads of GLiNER models on Hugging Face.
- 5,200+ GitHub stars
- 1.1 billion+ end users
- Trusted by leading AI teams building production AI

## Why small language models matter

Small language models are optimized for agentic workflows requiring numerous low-latency models specialized for specific tasks — model routing, tool calling, intent classification, response validation, and structured data extraction. Fine-tuned SLMs can match or outperform frontier models like GPT-4o on specific tasks at a fraction of the cost and latency. Fastino's mission is to make specialized small model fine-tuning accessible to every developer.

## Pages

- [Company homepage](https://fastino.ai)
- [Research blog](https://fastino.ai/blog)
- [Models hub](https://fastino.ai/models)
- [Collaboration with NVIDIA on fine-tuning Nemotron 3.5 Lightning models](https://fastino.ai/nvidia-collaboration)
- [Hugging Face hub](https://huggingface.co/fastino)
- [GLiNER2 GitHub](https://github.com/fastino-ai/GLiNER2)
- [Careers](https://jobs.ashbyhq.com/fastino-ai)

## Contact & community

- [LinkedIn](https://www.linkedin.com/company/fastinoai/)
- [X/Twitter](https://x.com/fastinoAI)
- [Discord](https://discord.com/invite/fastino)
- [YouTube](https://www.youtube.com/@Fastino-Labs)
- [Reddit (GLiNER community)](https://www.reddit.com/r/GLiNER/)
- Support: [support@fastino.ai](mailto:support@fastino.ai)