What gunpowder did to the knight, open tools are doing to drug discovery

What gunpowder did to the knight, open tools are doing to drug discovery

The medieval battlefield belonged to the knight. Armour, a warhorse, a lifetime of training — only the few who could afford all three ever stood on it.

What ended that was not a better knight. It was gunpowder. A gun in a farmhand's hands made generations of inherited skill beside the point. The knights did not disappear; what changed was who was allowed on the field.

Drug discovery has had the same shape. Drawing a molecule is the easy half. Knowing whether it is worth anything is the hard half, and that tool belonged to whoever owned the laboratory and the compute. Everyone else could have the idea and no way to test it.

We spent years building that judgement

Verifying PharmaOS took months on IBM's Heron quantum processor.

And we did not ask anyone to take our word for it.

First, we competed on someone else's leaderboard. On Polaris Hub, the benchmark for AI drug-property prediction, our engine holds first place in 16 ADME and toxicity tracks. The most recent came last week: predicting whether a compound crosses the cell membrane (MDR1-MDCK efflux ratio) from structure alone — correlation 0.802, against 0.725 for second place.

Second, we put a molecule through. A candidate from this engine has been filed as a composition-of-matter patent and taken into preclinical work. The experimental results came back inside the error bars of the prediction.

That engine is what scores your molecules here.

And the method is the same: approved drugs and inert compounds sit in the same table as your entries. We are not asking you to trust the score. You can check it.

The field is open right now

Two seasons are running.

Season 1 — Malaria. Target: PfDHODH in Plasmodium falciparum. Counter-target: human DHODH. Closes 30 September 2026. Prize $1,000.

Season 2 — Tuberculosis. Target: InhA in Mycobacterium tuberculosis. Counter-target: human FASN (ER domain). Closes 31 October 2026. Prize $2,000.

You submit a molecule as SMILES or InChI. Use Claude, GPT, Gemini, Qwen, KIMI, DeepSeek, or your own model — the challenge does not care how you arrived at the structure, only what the structure is.

Scoring is 100 points across six axes: activity, binding, selectivity, ADMET, novelty and synthesizability. The weights differ by season. Season 1 leans on activity. Season 2 splits weight toward binding and selectivity, because an InhA inhibitor that also hits the human enzyme is not useful to anyone.

That counter-target is the part worth understanding. It is easy to score well on binds the pathogen enzyme. It is much harder to bind the pathogen enzyme and leave the human homolog alone. Every season carries one.

If you would rather not publish your structure, you do not have to. Private submissions appear on the board with a masked SMILES.

The leaderboard is also an API — /api/seasons and /api/leaderboard?season=N return JSON, so you can pull standings, weights and per-axis breakdowns yourself.

One thing we will not overstate

A high score is a research hypothesis, not a drug. Every number here comes from computational scoring. It is not clinical validation, and it does not become one by ranking well. The anchors are in the table so you can see the scale for yourself rather than take our word for the ceiling.

Tools spread, and the field changes. That is what gunpowder did.

Challenge: https://huggingface.co/spaces/FINAL-Bench/open-discovery-challenge