Anthropic Launches Claude Science Beta: Multi-Agent Environment for Genomics and Proteomics

Anthropic launched Claude Science Beta on June 30, 2026 — a multi-agent platform for reproducible computational research in genomics, proteomics, and cheminformatics.

What is Claude Science

The platform is not a separate language model — it operates on top of existing Claude models and organizes their interaction into a specialized multi-agent environment for biomedical pipelines.

The central element is a coordinating agent. Upon receiving a task, it breaks it down into subtasks and delegates to specialized domain agents — in genomics, proteomics, or cheminformatics. In parallel, a separate reviewer agent verifies and corrects numerical data, references, and citations before the result leaves the platform.

• Beta launch — June 30, 2026
• Operates on top of existing Claude models, not a separate model
• Coordinating agent + specialized domain agents
• Separate reviewer agent for verifying numbers and citations
• Connected to more than 60 scientific databases
• Integration with NVIDIA BioNeMo capabilities

The target disciplines — genomics, proteomics, and cheminformatics — are unified by massive data volumes and multi-step computational pipelines. In such pipelines, reproducibility is critical: each intermediate step affects the final result, and manual documentation quickly falls behind the actual computational progress.

Each generated chart is accompanied by precise code, a description of the computational environment, and a complete history of agent messages. Any researcher or reviewer can reproduce the entire path from raw data to final visualization.

How Claude Science manages computations

The platform supports three types of computational resources within a single workflow: the researcher's local machine, a high-performance HPC cluster via SSH, and the cloud service Modal. Task distribution across resources occurs automatically, without manual data transfer.

For biomedical computations, access to current data is critical. The platform connects directly to more than 60 databases and uses NVIDIA BioNeMo capabilities — specialized models for molecular biology tasks. This makes it possible to build pipelines without manual imports of genomic annotations, protein structures, or chemical libraries at each stage.

Why is a separate reviewer agent needed?

Language models regularly make mistakes in details — numbers, units of measurement, gene identifiers, author names. In a scientific context, such an error is not merely a stylistic glitch, but potentially an incorrect conclusion that is then cited by other works.

Separating review into a distinct agent layer makes verification transparent: the researcher sees not only the final result, but also what was corrected and why. The public history of agent messages simplifies auditing for journal reviewers.

What this means

Claude Science is Anthropic's first explicit attempt to adapt multi-agent architecture to reproducible biomedical science tasks. The platform attacks the reproducibility problem through technical means: automatic capture of the entire computational chain in a structured, executable form. If the beta proves valuable, this could change the standards for publishing computational results — from "methods and materials" to an executable journal with a history of agent decisions.

Frequently Asked Questions

What computational environments does Claude Science support?

The platform manages computations on three types of resources: local machines, HPC clusters via SSH, and the cloud service Modal — within a single workflow without manual switching.

Which databases does the platform work with?

In the beta version, Claude Science is connected to more than 60 scientific databases and uses NVIDIA BioNeMo capabilities — specialized biomedical models for molecular biology tasks.