BrowseComp-Plus, perhaps the hardest popular deep research task, is now solved at nearly 90%...
... and all it took was a 150M model ✨
Thrilled to announce that Reason-ModernColBERT did it again and outperform all models (including models 54× bigger) on all metrics Model:
If you are not familiar with the model, it is a multi-vector model built at @LightOnIO optimized for reasoning-intensive retrieval, outperforming model 45x bigger on the BRIGHT benchmark

And now, it does it again, for agentic search/deep research huggingface.co/lightonai/Reas…

Your coding agent is burning tokens on grep like it's 1973
Because semantic search means remote APIs & babysitting an index
Introducing ColGrep & LateOn-Code
SOTA code retrieval with lightweight models.
Wins 70% vs grep. 15.7% less tokens.
Local, open & free. Runs on a toaster. Blog post: huggingface.co/blog/lightonai…
HF collection: huggingface.co/collections/li…
Start saving token today: github.com/lightonai/next…

We release everything, the models, the training data, the recipes and most importantly, ColGrep, the harness to enhance your agent and save tokens

Agentic RAG is the✨hot✨new✨thing
That is why I am thrilled to announce that @LightOnIO is releasing Reason-ModernColBERT, reaching the top of the popular BRIGHT benchmark and outperforming models more than 45 times its size on reasoning-intensive retrieval! Model:
So what is reasoning-intensive retrieval?
As explained in the BRIGHT paper, most existing retrieval tasks leverage lexical or semantic-based retrieval, but in practice, some tasks can require reasoning to determine what the target relevancy is huggingface.co/lightonai/Reas…

Among all those LLM releases, here is an important retrieval release:
To overcome limitations of awesome ModernBERT-based dense models, today @LightOnIO is releasing GTE-ModernColBERT, the very first state-of-the-art late-interaction (multi-vectors) model trained using PyLate🚀 Model link: huggingface.co/lightonai/GTE-…
GTE-ModernColBERT is trained on top of the GTE-ColBERT model using knowledge distillation on the MS MARCO dataset and is the first SOTA model trained using PyLate!
Get started with PyLate using the documentation:
lightonai.github.io/pylate/