.svg)
Research And Development
R&D Overview
.avif)
Advancing Generative AI through Innovation
The R&D team at LightOn plays a pivotal role in advancing the field of generative AI through continuous innovation and development. Their expertise spans across creating and fine-tuning large language models (LLMs) that form the backbone of the Paradigm platform, a comprehensive AI solution designed for enterprise use. This platform simplifies the integration of generative AI into business workflows, offering both on-premise and cloud options to ensure flexibility and scalability for various business needs.
r&d publicationsRecent R&D Posts
Adaptive Chunking: Reasoning Starts Before the LLM Sees a Token
Document-aware chunking selection for production RAG systems
CTA Title
Lorem Ipsum
Deep Research is now Open
Agent-ModernColBERT adds ~10% over Reason-ModernColBERT on BrowseComp-Plus, stays at 149M parameters, and brings GPT-5 + Qwen3-8B-level retrieval performance to a fully open stack.
CTA Title
Lorem Ipsum
🔴 The Retriever You Actually Need
Introducing LateOn and DenseOn, two Apache 2.0 retrievers: SOTA on BEIR, built to generalize.
CTA Title
Lorem Ipsum
Document Intelligence at First Sight
OriOn-Qwen-SR1: Fast Implicit Reasoning for Long Documents
CTA Title
Lorem Ipsum
Open-Source LightOnOCR-2 Just Outscored Claude, GPT-5, Qwen3, Mistral and Mathpix at Table Extraction
The most valuable information in enterprise documents doesn't live in paragraphs. It lives in tables
CTA Title
Lorem Ipsum
The Bloated Retriever Era Is Over
Reason-ModernColBERT tops BrowseComp-Plus, the most rigorous agentic search benchmark, across every metric, with 54× fewer parameters and fewer search calls.
CTA Title
Lorem Ipsum
NOVA: A Guide to Actually Measuring How Your Agent Works on Your Data
Numbers Over Vibes: Building a RAG Evaluation Framework That Actually Works
CTA Title
Lorem Ipsum
