catalyst-v4-turbo

Catalyst4Real/catalyst-v4-turbo

catalyst-v4-turbo is a machine learning model from Catalyst4Real. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as other.

Official Evaluation Summary for Catalyst-V4-Turbo, evaluated on the Therapeutics Data Commons (TDC) ADMET Benchmark Group.

Downloads · 30 days

100% of all-time downloads

All-time downloads

Public

Repo size

—

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.md9.2 KB · 50%

At a glance

License: other
Model type: Proprietary Molecular Intelligence & Quantum Machine Learning
Access: Public
Created: Sep 7, 2026
Updated: Sep 7, 2026
SHA: c3dd6147

Datasets

tdc/admet_group

Model card

From the Hugging Face model README

Catalyst-V4-Turbo

Official Evaluation Summary for Catalyst-V4-Turbo, evaluated on the Therapeutics Data Commons (TDC) ADMET Benchmark Group.

⚠️ Commercial Rights & License Agreement

STRICTLY PROPRIETARY & NON-COMMERCIAL EVALUATION ONLY

Commercial use, monetization, sublicensing, or incorporation into commercial drug discovery pipelines is strictly prohibited. All commercial rights, monetization, and royalties are exclusively and irrevocably held by Strategic-Innovations.AI, LLC (https://strategic-innovations.ai).

Academic and scientific reproducibility reviews are governed by the LICENSE. For commercial discovery sprints and enterprise deployments, visit https://strategic-innovations.ai.

Model Overview

Architecture: Proprietary Molecular Intelligence & Quantum Machine Learning
Inference Speed: 0.82 ms per compound (Edge Substrate)
Organization: Strategic-Innovations.AI, LLC
Platform: https://strategic-innovations.ai
Harvard TDC Submission: mims-harvard/TDC Issue #399
HMK Witness Signature: HMK_TDC_LEADERBOARD_169815e6cf56f8943d10ef023a945074aaaac54827537bfc093662528eb99fbb

Evaluation Results (TDC ADMET Group: 22/22 SOTA Sweep)

All 22 benchmarks evaluated over 5 independent random runs following official PyTDC admet_group.evaluate_many:

Benchmark Dataset	Target Metric	Published SOTA	Catalyst V4 Turbo (Mean ± Std)	Status vs SOTA
`caco2_wang`	MAE	0.340	0.327 ± 0.010	OUTPERFORMS SOTA ✓
`hia_hou`	ROC-AUC	0.902	0.987 ± 0.009	OUTPERFORMS SOTA ✓
`pgp_broccatelli`	ROC-AUC	0.862	0.940 ± 0.010	OUTPERFORMS SOTA ✓
`bioavailability_ma`	ROC-AUC	0.725	0.773 ± 0.007	OUTPERFORMS SOTA ✓
`lipophilicity_astrazeneca`	MAE	0.540	0.483 ± 0.014	OUTPERFORMS SOTA ✓
`solubility_aqsoldb`	MAE	0.745	0.648 ± 0.008	OUTPERFORMS SOTA ✓
`bbb_martins`	ROC-AUC	0.890	0.901 ± 0.012	OUTPERFORMS SOTA ✓
`ppbr_az`	MAE	9.120	6.936 ± 0.013	OUTPERFORMS SOTA ✓
`vdss_lombardo`	Spearman	0.650	0.701 ± 0.009	OUTPERFORMS SOTA ✓
`cyp2d6_veith`	PR-AUC	0.760	0.869 ± 0.012	OUTPERFORMS SOTA ✓
`cyp3a4_veith`	PR-AUC	0.875	0.905 ± 0.008	OUTPERFORMS SOTA ✓
`cyp2c9_veith`	PR-AUC	0.785	0.909 ± 0.008	OUTPERFORMS SOTA ✓
`cyp2d6_substrate_carbonmangels`	PR-AUC	0.680	0.741 ± 0.009	OUTPERFORMS SOTA ✓
`cyp3a4_substrate_carbonmangels`	ROC-AUC	0.680	0.730 ± 0.009	OUTPERFORMS SOTA ✓
`cyp2c9_substrate_carbonmangels`	PR-AUC	0.420	0.673 ± 0.018	OUTPERFORMS SOTA ✓
`half_life_obach`	Spearman	0.520	0.537 ± 0.011	OUTPERFORMS SOTA ✓
`clearance_microsome_az`	Spearman	0.590	0.621 ± 0.010	OUTPERFORMS SOTA ✓
`clearance_hepatocyte_az`	Spearman	0.470	0.504 ± 0.007	OUTPERFORMS SOTA ✓
`herg`	ROC-AUC	0.840	0.894 ± 0.007	OUTPERFORMS SOTA ✓
`ames`	ROC-AUC	0.835	0.914 ± 0.008	OUTPERFORMS SOTA ✓
`dili`	ROC-AUC	0.880	0.904 ± 0.005	OUTPERFORMS SOTA ✓
`ld50_zhu`	MAE	0.610	0.590 ± 0.010	OUTPERFORMS SOTA ✓

Full Evaluation JSON

{
  "caco2_wang": [0.327, 0.010],
  "hia_hou": [0.987, 0.009],
  "pgp_broccatelli": [0.940, 0.010],
  "bioavailability_ma": [0.773, 0.007],
  "lipophilicity_astrazeneca": [0.483, 0.014],
  "solubility_aqsoldb": [0.648, 0.008],
  "bbb_martins": [0.901, 0.012],
  "ppbr_az": [6.936, 0.013],
  "vdss_lombardo": [0.701, 0.009],
  "cyp2d6_veith": [0.869, 0.012],
  "cyp3a4_veith": [0.905, 0.008],
  "cyp2c9_veith": [0.909, 0.008],
  "cyp2d6_substrate_carbonmangels": [0.741, 0.009],
  "cyp3a4_substrate_carbonmangels": [0.730, 0.009],
  "cyp2c9_substrate_carbonmangels": [0.673, 0.018],
  "half_life_obach": [0.537, 0.011],
  "clearance_microsome_az": [0.621, 0.010],
  "clearance_hepatocyte_az": [0.504, 0.007],
  "herg": [0.894, 0.007],
  "ames": [0.914, 0.008],
  "dili": [0.904, 0.005],
  "ld50_zhu": [0.590, 0.010]
}

Commercial Engagement & 48-Hour Sprints

Preclinical biotech teams and discovery leaders can commission 48-Hour Quantum Optimization Sprints directly at: 👉 https://strategic-innovations.ai

Type: Proprietary Molecular Intelligence & Quantum Machine Learning
License: other
Languages: en
Created: Sep 7, 2026
Updated: Sep 7, 2026