Drug discovery has long been bottlenecked not by hits, but by development. Latent-X2 is our step toward designing the right molecule from the start.
Technical report: tinyurl.com/latent-X2-re...
Blog: www.latentlabs.com/latent-x2
Apply for access: partnerships@latentlabs.com
To enable reproduction, we're publishing the sequences and structures of a lab-validated antibody design for each target — accessible on the Latent Labs Platform without sign-in.
For the first time: immunogenicity assessment of an AI-generated antibody. De novo VHH binders cleared ex vivo assays across a ten-donor human panel.
Combined with developability profiles matching approved therapeutics — drug-like from the first generation.
Latent-X2 achieves picomolar to nanomolar affinities across VHH, scFv, and macrocyclic peptides — testing just 4-24 designs per experiment.
Hits on K-Ras, long considered undruggable, matching trillion-scale screens while testing 11 orders of magnitude fewer sequences.
5 months from Latent-X1 to Latent-X2. AI-generated antibodies with drug-like developability and low immunogenicity in human panels — zero-shot.
Available now for selected partners.
Drug discovery has long been bottlenecked not by hits, but by development. Latent-X2 is our step toward designing the right molecule from the start.
Technical report: tinyurl.com/latent-X2-re...
Blog: www.latentlabs.com/latent-x2
Apply for access: partnerships@latentlabs.com
To enable reproduction, we're publishing the sequences and structures of a lab-validated antibody design for each target — accessible on the Latent Labs Platform without sign-in.
For the first time: immunogenicity assessment of an AI-generated antibody. De novo VHH binders cleared ex vivo assays across a ten-donor human panel.
Combined with developability profiles matching approved therapeutics — drug-like from the first generation.
The AWS partnership has been providing scalable compute through Amazon SageMaker HyperPod and an incredible community that's accelerated our mission.
Instead of waiting weeks or months for lab experiments, researchers can now get to precision breakthrough molecules at the push of a button.
Our Latent-X model uses AI to design proteins and model molecular interactions – moving beyond what nature gives us to precisely engineering what we need for therapeutics, food tech, and climate solutions.
🧬 From observational science to engineering – this is the transformation we're driving in biology.
Latent-X has been available for over two weeks and we couldn't be more excited about the science already being done with it.
Our CEO @saakohl.bsky.social joined @bloomberg.com News Daybreak Europe today to talk why we're building it, and the impact it is having.
Watch here: tinyurl.com/yjzyhr23
🔬 Mini-binders represent a new therapeutic modality — for targeted delivery, diagnostics, and therapeutic inhibitors.
For partnerships: partnerships@latentlabs.com
🚀 Available now on the Latent Labs Platform with free tier access:
* No-code interface
* Upload targets, select epitopes, generate binders
* 10x faster than previous methods
* Computational ranking for lab prioritization
🧪 Lab-validated target specificity across all tested mini-binders.
See the all-against-all binding screen via our in-house mDisplay assay below:
⚡ Watch Latent-X generate a picomolar mini-binders from scratch — solving the geometric puzzle of binding at the all-atom level in real-time.
Every atom placed with precision to create the biochemistry required for tight, specific binding.
🔬 Look at the generated non-covalent bonds — hydrogen bonds, pi-stacking, van der Waals interactions.
Latent-X doesn't just predict binding; it generates the exact biochemical interactions needed for high affinity and specificity.
🎯 Extensive lab validation shows leading performance:
* 100% target success rate (5/5 therapeutically relevant proteins)
* Picomolar binding affinities achieved
* 10-64% experimental hit rates
* Testing just ~100 designs per target vs millions traditionally
🧬Latent-X is our first frontier model for protein design.
For mini-binders, our AI model achieved picomolar binding affinities — the strongest reported vs prior methods in head-to-head lab validation.
Explore our best binders here: platform.latentlabs.com
More details on the breakthrough in this🧵
In extensive wet lab experiments, Latent-X achieved what would typically require millions of candidates by testing as little as 30-100 candidates per target—yet achieved strong binding affinities down to the picomolar ranges.
Sign-up for early access at: platform.latentlabs.com
‘Latent Labs launches web-based AI model to democratize protein design.’
Thank you @techcrunch.com and Marina Temkin for the coverage.
Read the full story: tinyurl.com/Latent-X-Tec...
Starting with macrocycles and mini-binders, expanding to nanobodies and more. Our mission: make biology programmable to make drug design instantaneous.
Join early access: platform.latentlabs.com
Technical report: tinyurl.com/latent-X
Technical details: www.latentlabs.com/latent-x
Latent-X solves the geometric puzzle of binding at the all-atom level, generating novel binders from scratch while obeying nature's biochemical rules.
Scientists can now access lab-validated AI workflows without needing AI expertise or infrastructure.
Traditional drug discovery screens millions of random molecules with <1% hit rates. Latent-X flips this: we generate 30-100 precise designs per target and achieve lab success rates up to 100%.
The efficiency gains are dramatic.
5 months from funding to shipping our first frontier model. Latent-X achieves state-of-the-art hit rates for macrocycles and mini-binders, with picomolar binding affinities — a breakthrough in de novo protein binder design.
Available now on our no-code platform!
