Technology
Distill Labs
The developer platform for building task-specific Small Language Models (SLMs): achieve LLM-level accuracy in hours with minimal data via knowledge distillation.
Distil Labs is the developer platform for deploying custom Small Language Models (SLMs) fast. We automate the entire ML pipeline, using knowledge distillation to fine-tune a compact SLM with just a prompt and a few dozen examples (no manual labeling required). This process delivers state-of-the-art accuracy, with models up to 400x smaller than their teacher LLMs. For example, our 1B PII redaction model achieved 0.81 $\pm$ 0.02, effectively matching a 600B+ class LLM like DeepSeek 3.1. Ship your custom model in under eight hours for private, on-device deployment across use cases like text classification, Open Book QA, and structured data extraction.
What builders pair with Distill Labs
Projects using both technologies. Select a pairing to see a project.
Pairing: Hugging Face
Teaching small language models a thing or two
Pairing: Python
Teaching small language models a thing or two
Pairing: Transformers
Teaching small language models a thing or two
Recent Talks & Demos
Showing 1-1 of 1
