Intology Launches Locus: Autonomous Research Agent Outperforms GPT-6 and Claude 5.5 in Molecular Design

1 reports archived1 independent sourcesupdated 11/4/2026, 01:05:23

Synthesis & Latest Updates

1 Sources Cross-Validated

Intology has introduced Locus, an automated research system capable of autonomously proposing experiments, running parallel batches on GPUs, and iterating on successful models. In molecular prediction and design tasks, Locus outperformed agents built on GPT-6 Astra and Claude Opus 5.5, demonstrating superior generalization and sample efficiency in proprietary data scenarios.

LATEST/Intology has introduced Locus, an automated research system capable of autonomously proposing experiments, running parallel batches on GPUs, and iterating on successful models. In molecular prediction and design tasks, Locus outperformed agents built on GPT-6 Astra and Claude Opus 5.5, demonstrating superior generalization and sample efficiency in proprietary data scenarios.

Hourly heat curve

2 fully observed hours

Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first

Rohan Paul (@rohanpaul_ai)T2Nov 4, 01:05·78 pts
Intology Launches Locus: Autonomous Research Agent Outperforms GPT-6 and Claude 5.5 in Molecular Design
- Locus achieves a fully automated research loop from hypothesis generation to experiment execution and model iteration.
- In molecular scoring, small-molecule potency prediction, and antibody design workflows, Locus outperforms existing agent solutions relying on frontier LLMs.
- The system demonstrates higher sample efficiency and solution diversity in limited-data and out-of-distribution tasks.
View Related Reports & Analysis Original Source