Featured

Introducing LightOn Console

Production-grade retrieval pipeline. Free Tier. Sign up. Ship

June 2, 2026

Read post

mDenseOn and mLateOn: more signal, less noise for multilingual agentic search

Two 307M-parameter multilingual retrieval models that achieve state-of-the-art results on English general-domain retrieval (BEIR), long-document retrieval (MLDR), multilingual retrieval (MIRACL) and code retrieval (MTEB Code).

July 30, 2026

Read post

Introducing LightOn-rerank: SOTA multimodal LLM reranker

One 2-billion-parameter model reads text passages and scanned pages on the same relevance scale, from a single adapter and a single deployment.

July 16, 2026

Read post

Faceted search has entered the agentic era

Meet LightOn Facets, the simplest way to find documents by what they are, not just what they say.

June 26, 2026

Read post

Fewer Tokens, Better Answers: Give Your Agent Search, Not Raw Files

Same model, same questions. Search cut tokens 36% and won 75% of blind comparisons.

June 24, 2026

Read post

How HR due diligence saved an M&A divestiture

The impossible use case: Season 1, Episode 7: HR

June 23, 2026

Read post

Multi-vector retrieval now on the infrastructure you already run

ColBERT-grade retrieval quality on the infrastructure teams already operate, at a fraction of the storage and serving cost.

June 16, 2026

Read post

LightOn Demonstrates the Flexibility of Its OCR Model by Adapting It to Arabic Through Targeted Training

LightOn demonstrates the flexibility of LightOnOCR-2, its document understanding model, by adapting it to Arabic through fine-tuning.

June 12, 2026

Read post

Systematic Grounding: How to Control AI Agent Token Costs

Token prices are collapsing. Your bill is climbing anyway. The agent era rewrote the math, and most teams now pay full price for tokens their model never needed.

June 11, 2026

Read post

The At-Risk Supplier You Were About to Double Orders With

Impossible Use Cases - Season 1, Episode 06: Quality

June 10, 2026

Blog

Introducing LightOn Console

mDenseOn and mLateOn: more signal, less noise for multilingual agentic search

Introducing LightOn-rerank: SOTA multimodal LLM reranker

Faceted search has entered the agentic era

Fewer Tokens, Better Answers: Give Your Agent Search, Not Raw Files

How HR due diligence saved an M&A divestiture

Multi-vector retrieval now on the infrastructure you already run

LightOn Demonstrates the Flexibility of Its OCR Model by Adapting It to Arabic Through Targeted Training

Systematic Grounding: How to Control AI Agent Token Costs

The At-Risk Supplier You Were About to Double Orders With

No matching results found