Free ADMET Prediction Online
Run ADMET, mutagenicity, and pKa predictions on any small molecule. Absorption, distribution, metabolism, excretion, and toxicity in one view — no login, no API key, no rate limits for humans.
What you get on every compound page
Absorption
Human intestinal absorption, P-gp substrate / inhibitor status, oral bioavailability indicators, and Caco-2 permeability flags.
Distribution
Blood-brain barrier penetration, plasma protein binding, volume of distribution estimates, and CNS-active indicators.
Metabolism
CYP450 substrate and inhibitor predictions across CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 — the isoforms that matter for drug-drug interaction risk.
Excretion
Total clearance estimates and renal OCT2 substrate flags.
Toxicity & mutagenicity
AMES mutagenicity (Salmonella reverse mutation), hERG cardiotoxicity, hepatotoxicity, skin sensitization, and acute oral toxicity (LD50). Predictions tagged with uncertainty so you know when to trust them.
pKa
Site-specific pKa predictions for ionizable groups, with explicit acid/base classification — useful for solubility, formulation, and salt-form selection.
Honest comparison with what else is out there
| Feature | OCSR.ai | SwissADME | pkCSM | ProTox-II | admetSAR |
|---|---|---|---|---|---|
| Free | ✓ Free | ✓ Free | ✓ Free | ✓ Free | ✓ Free |
| No login | ✓ No login | ✓ | ✓ | ✓ | ✓ |
| Paste image → predict | ✓ Yes | SMILES only | SMILES only | SMILES only | SMILES only |
| Mutagenicity (AMES) | ✓ | — | ✓ | ✓ | ✓ |
| pKa site prediction | ✓ Per-site | — | — | — | — |
| Compound search engine | 1.3T molecules | — | — | — | — |
| Similarity search | ✓ Tanimoto | — | — | — | — |
| PDF report export | ✓ One click | Manual | Manual | Manual | Manual |
| Per-compound AI chat | ✓ Darvi | — | — | — | — |
SwissADME, pkCSM, ProTox-II, and admetSAR are great academic tools. The honest gap: they cover one slice (drug-likeness, or toxicity, or oral bioavailability) and stop there. OCSR.ai gives you all of it on the same compound page, plus search, similarity, and a PDF report.
How to run an ADMET prediction
- Find your compound. Use the search bar on the homepage — type a name (e.g. “atorvastatin”), paste a SMILES, drop an image, or upload a MOL file.
- Open the compound page. Predictions are already computed and visible — no “run prediction” button to click. Examples: atorvastatin · imatinib · metformin.
- Scroll to ADMET, Toxicity, and pKa tabs. Each prediction shows the value, a qualitative interpretation, and an uncertainty indicator so you know when to be skeptical.
- Export the PDF. The PDF includes structure, identifiers, predictions, and references — formatted to drop into a regulatory filing, a thesis, or a med-chem review meeting.
Frequently asked questions
How reliable are the predictions?
All predictions are model outputs, not ground truth. We display uncertainty metrics so you can judge confidence. Use them for triage and prioritization — not to replace wet-lab validation for safety-critical decisions. The same caveat applies to every ADMET tool on the market.
Can I use this for commercial drug discovery?
Yes. The public site is free for individual research use. For commercial pipelines, programmatic access, or bulk runs, write to [email protected].
What models power these predictions?
We use a stack of graph neural networks and physics-informed regressors trained on large open ADMET datasets (Therapeutics Data Commons, ChEMBL, public PK studies). The exact models and training procedure are documented in our about page.
Get ADMET in 5 seconds
Search any compound by name, SMILES, or image. ADMET, toxicity, mutagenicity, and pKa are pre-computed and waiting on the page. No queue, no signup.
Open OCSR.ai →