agentic discovery needs
data, models, wet-lab execution.
We screen entire proteomes across hundreds of millions of molecules in the lab to power your ML models and your initial new medicine discoveries.
350M molecules tested per protein.
score billions of molecules with trained or finetuned models.
Generated Data in weeks; provider-backed Molecule Fulfillment workflows.
LULA-1 & LULA-1.5
Score a million molecules for $10.
What it is
LULA-1 predicts how strongly any molecule binds your target.
What it needs
Just a SMILES string and the protein's amino-acid sequence. No crystal structure, no docking, no folding.
How it's trained
500M real wet-lab binding measurements. Ligand data only — no structural data, ever.
What you get
A binding score you can use to rank millions of molecules in minutes.
Structure-based methods like Boltz-2 have to fold and dock a candidate before they can score it. LULA-1 skips that step entirely.
LULA-1
FastScreens broadly across many targets in minutes — the default for scoring millions of molecules against a target.
LULA-1.5
High-ResApplied selectively on the hardest targets — protein-protein interfaces, flat pockets — where precision matters most.
5,000× cheaper per molecule
~7 min vs ~4 years for fold-then-dock methods to score 1M molecules
0.83 AUROC on novel targets
why data generation
We screen entire proteomes against hundreds of millions of molecules, to power proteome-wide interaction models like LULA-1 and LULA-1.5.
The technology
Pico-scale library × library screening: every molecule in one library tested against every protein in another, simultaneously.
The economics
Multiplexed at massive scale instead of one compound at a time — orders of magnitude cheaper than the status quo.
Feeds the next model
Every screen becomes training signal — sharpening every future LULA-1 and LULA-1.5 run.
300M data points per protein
1,000 proteins a month capacity
~900B data points a quarter
benchmark results, verified
Real signal, not memorization.
Mean global AUROC is 0.744 across two private validation panels and 0.873 across two public test panels. With labels randomized, AUROC returns to chance at 0.500.
Private average
2 validation panels
Public average
2 test panels
Random labels
chance control
Unweighted mean of panel-level global AUROC. Private: full validation (0.739) and sequence-overlap validation (0.748). Public: nonoverlap test (0.858) and overlap test (0.887).
fast, then precise where it counts
LULA-1.5 turns hard targets into real hits.
Top-1000 precision on a 12-target product panel, fixed LULA-1.5 checkpoint. LULA-1 covers every target broadly; LULA-1.5 is the selective high-res pass for the hardest ones — protein-protein interfaces, flat pockets — not a universal upgrade.
attention discovery
Attention finds real contacts across 164 independent structures.
LULA-1 trains on ligand binding data and protein sequence alone. This is what its attention finds anyway, checked against real, recent PDB structures it never trained on.
- 01
Trained on binding data alone
LULA-1 learns from real ligand binding measurements and protein sequence. No 3D structure is part of training, ever.
- 02
Attention ranks the residues
Its cross-attention between ligand and protein surfaces which residues correlate most with binding — a sequence-level signal alone.
- 03
Structure models seed a pocket
Those ranked residues, together with the ligand, seed a predicted pocket and binding pose using Chai-1 and other structure models.
- 04
Checked against real structures
The result correlates closely with real, independently solved PDB structures — on targets the model never trained on.
500M training data points
8% of the human proteome
300M molecules screened per protein
discovery pipeline
Data Generation for your target. A model you can use. Molecules you can validate in under a week.
Every order is exclusive to you — private data on your protein, not a shared pool. Get a model fine-tuned on it to power your own discovery program, and keep fine-tuning LULA-1 and LULA-1.5 to go faster as you generate more.
- 01
Disease target
bring the target
- 02
Wet-lab data
generate data at scale
- 03
Trained models
score the molecule space
- 04
Molecule Fulfillment
quote provider-backed fulfillment
screen → fulfill, in one flow
10M OnePot molecules to your bench.
OnePot is an early partner helping Om get molecules to customers fast: score any vendor’s molecule space with LULA-1, then order agentically and have the winners delivered.
Real SDK call
job = client.hub.lula1(
protein_sequence="YOUR_TARGET_PROTEIN_SEQUENCE",
vendor="one-pot",
n=10_000_000,
threshold=0.3,
top_k=10_000,
job_name="lula1-onepot-10m",
)- 1
Score 10M accessible OnePot molecules against your target with LULA-1.
- 2
Get back the top-10K ranked hits by predicted binding score.
- 3
Quote provider-backed Molecule Fulfillment on those exact top hits.
- 4
Checkout or invoice, then the order moves through fulfillment and delivery.
Early fulfillment partner
callable by agents
Connect Om where the work is happening.
Use Om from Codex or Claude Code with hosted MCP OAuth. Direct API and CLI paths stay available when you need production orchestration.
Codex
OAuth remote MCP
Add the hosted Om MCP endpoint, then complete the OAuth login.
Setup
codex mcp add omtx --url https://agents.omtx.ai/mcp
codex mcp login --scopes email omtxRestart Codex, then ask for om_status or pricing_get to confirm the server is attached.
Claude Code
OAuth remote MCP
Register the remote HTTP MCP server, then authenticate from /mcp.
Setup
claude mcp add --transport http omtx https://agents.omtx.ai/mcpRestart Claude Code, run /mcp, complete OAuth in the browser, then ask for om_status.
Latest Articles
Vibe Discovery Is Coming
Om MCP brings Diligence, Data Access, Artifacts, Jobs, and Hub workflows into Codex, Claude Code, and OAuth-capable MCP harnesses.
April 8, 2026 · 12 min read
Read More
Data Data Everywhere
Stream qualifying Generated Data through the Om API after Data Generation output is available to your account.
May 1, 2025 · 5 min read
Read More
Om Launches Diligence
Diligence pairs our proteome-scale research assistant with pay-as-you-go Om API workflows, so you can evaluate targets in the UI or programmatically.
April 30, 2025 · 6 min read
Read MoreWhile you're vibin'...
Watch the discovery playlist.
A little birdy said to ask Om MCP to make a vibe video for your discovery efforts. Want to ask for your favorite vibe video style?