+30% avg
accuracy lift on classification & extraction tasks vs. base Llama 3.1
~7 days
until your first auto-improvement run lands in production
0 lines
of fine-tuning code you have to write, ever
$0/retrain
starting price — pay for inference, the improvement is included
RESEARCH
From our research lab.
Research
GLiNER2-PII: Open Source Privacy Filtering with PII Detection
We're releasing GLiNER2-PII, a 300M parameter SOTA open-source model for detecting and redacting PII.
Research
Introducing GLiGuard
16x faster safety moderation with a small language model.
Research
The Agent Behind Pioneer
How fine-tuning small language models creates a closed-loop system.
Product
Introducing Pioneer
An agent designed to fine-tune and run open-source language models at scale.
Research
GLiNER2 for Agentic Extraction
Specialized architectures for targeted information extraction in agentic pipelines.
Research
GLiNER for Modern NER
Moving from universal models to domain-specific named entity recognition.