ISM Benchmarks: ADMET (Insilico)

28-endpoint ADMET benchmark suite on ScienceAIBench/InsilicoBench/DDB.

Composite

84.6

Experimental validation

Wet-lab confirmed

Stages

Lead ID / ADMETIND-enabling

Modalities

small-molecule

Task types

admet-regressionadmet-classification

Size

benchmarks: 28

License

CC-BY

First release

2025-02

Last updated

2026-04

Official site

→ project page

Leaderboard

→ leaderboard

Dataset

→ dataset

Code / GitHub

→ repository

HuggingFace

→ HF

Paper

ISM ADMET benchmark methodology · I, n, s, i, l, i, c, o, , A, I, , T, e, a, m · 2025 · paper · doi:N/A — portal · 7 citations

Flags

none

Experts

Alex Zhavoronkov, Alex Aliper

Groups

Insilico Medicine

Hosted by

ScienceAIBench, InsilicoBench, Drug Discovery Benchmarks (DDB)

Related benchmarks

TDC ADMET Group, ADMET-AI

Rubric (7-criterion)

rigor

coverage

maintenance

adoption

quality

accessibility

industry_relevance

Notes

Broader endpoint coverage than TDC ADMET. Side-by-side with TDC mirror on DDB.

← Back to all benchmarks

Compare:

Open comparison →