Apheris logo
Menu

ApherisFold

Run the latest co-folding models, such as OpenFold3 and Boltz-2, securely within your own environment — giving you full control over data, benchmarking, and fine-tuning.

Apply co-folding models to your drug discovery programs while keeping your data fully protected

Benchmark models on your in-house data


Explore outputs via a GUI or integrate them into existing workflows


Securely fine-tune models to your domains of interest

Access ApherisFold

Optimized for enterprise security and scalability requirements

Built for scientists, ApherisFold integrates seamlessly into existing R&D stack — supplementing, not replacing, your internal pipelines. It scales efficiently across multiple GPUs, users, and models, and meets enterprise-grade security and performance requirements.

Documentation

Secure local inference

Deploy ApherisFold locally such that no data leaves your corporate network and internal security policies are met.

In-house model benchmarking

Securely evaluate and compare models on your in-house data using performance guidelines and a benchmarking kit.

Model fine-tuning

Securely customize models to your domain of interests, either on your own data or on third-party data via federated data networks.

Visualization and integration

Explore molecular structures through 3D visualization via a scientist-friendly GUI or connect outputs to existing drug discovery workflows programmatically via an API.

Federated benchmarking in beta testing

Participating companies benchmark models locally on their proprietary data, with all computations staying inside their environments. Only aggregated, privacy-preserving performance metrics are shared to a joint dashboard, creating a federated view of model performance across organizations.

One application - multiple ways to use it

Connect outputs programmatically via an API

Explore 3D molecular visualization through a scientist-friendly GUI

Our users apply ApherisFold for the following drug discovery tasks

Apheris is the partner for secure local inference, benchmarking, and fine-tuning of OpenFold3

This AISB Network initiative aims to improve OpenFold3’s ability to predict complex molecular interactions relevant to industrial drug discovery. The model, developed by the AlQuraishi Lab at Columbia University, will be trained on participating members' proprietary protein–small molecule interaction data in a confidentiality-preserving federated environment powered by Apheris. Participating members are AbbVie, Astex, Bristol Myers Squibb, Johnson & Johnson, and Takeda.

Read the Press Release

Apheris powers the Federated OpenFold3 Initiative

Access ApherisFold

Securely use the latest co-folding models, like OpenFold3 and Boltz-2, on your proprietary data.

Click here if the form doesn't work