In the world of open-source AI agents, a new leader has emerged. As of May 10, 2026, Hermes Agent, developed by Nous Research, has taken the top spot on OpenRouter’s global daily rankings. It now outperforms OpenClaw in both usage volume and developer activity, marking a significant shift in the AI agent landscape.

The Rise of Hermes and Its Architectural Approach

Hermes Agent’s success is rooted in its unique design. Unlike OpenClaw, which centers around a web-based routing gateway connecting multiple messaging platforms, Hermes focuses on a “do, learn, improve” cycle. After completing a task, it reflects on its performance and creates reusable skill files to enhance future operations.

This self-improving loop allows Hermes to become more efficient the longer it runs. Its memory system combines a persistent identity snapshot, a full-text search database of past sessions, and procedural skills that capture task logic. This setup helps Hermes optimize workflows over time, making it highly adaptable for specific user needs.

Rapid Development and Feature Expansion

Since launching in February 2026, Hermes has been rolling out major updates at a fast pace. Its latest release, version 0.13.0, added features like a Kanban task board with monitoring tools, a command to lock the agent on a specific goal, and improved security measures. It now supports 20 messaging platforms, including Google Chat and QQBot.

The frequent updates have helped Hermes stay ahead in functionality and integration options. Support for new inference paths, native cloud platform compatibility, and security enhancements show a strong focus on usability and safety. This rapid development cycle has kept Hermes at the forefront of open-source AI agents.

Security Challenges in Large-Scale OpenClaw Deployments

While OpenClaw was once the dominant open-source agent, its large scale has come with security issues. In early 2026, multiple vulnerabilities were disclosed, with some scoring very high on security risk scales. Nearly a thousand suspicious entries were found in security audits, and thousands of instances were publicly exposed, raising concerns about its safety.

In comparison, Hermes’s security record is relatively shorter, but it also has some issues. Recent vulnerabilities included a missing authentication flaw in its webhook system. However, the Hermes team responded quickly by addressing these bugs and implementing better security practices with each update.

Migration and Future Outlook

For developers already using OpenClaw and interested in trying Hermes, the transition is designed to be smooth. Hermes can automatically import settings, memories, and API keys from existing installations. Tools are available to preview the migration process or selectively transfer data, making it easier to switch or run both agents in parallel.

The broader trend in the open-source community shows a split between two philosophies. One favors broad reach across many platforms, like OpenClaw, while the other emphasizes depth of learning and self-improvement, exemplified by Hermes. Hermes’s rise to the top indicates a growing interest in agents that get better over time and become more tailored to individual workflows.

Overall, Hermes’s new status as the most-used open-source AI agent reflects a shift towards smarter, self-optimizing tools. As the ecosystem evolves, developers will likely continue choosing between these two approaches based on their specific needs and security considerations.

Inspired by

• https://www.marktechpost.com/2026/05/10/openclaw-vs-hermes-agent-why-nous-researchs-self-improving-agent-now-leads-openrouters-global-rankings/

Sources

• github.com
• openrouter.ai
• steipete.me
• github.com
• github.com