Nous Research’s Hermes Agent processed 224 billion tokens on OpenRouter on May 10, surpassing OpenClaw’s 186 billion to claim the top position on the platform’s global daily inference rankings. It is the first time since OpenClaw’s November 2025 launch that a competing open-source agent has held the daily lead on what has become the most visible public leaderboard for AI agent usage.
The Numbers
The ranking flip is narrower than the headline suggests. According to Tech Times, OpenClaw still leads the cumulative all-time chart at 9.17 trillion tokens against Hermes’s 6.35 trillion. The daily figure is the leading indicator of where developers are routing new workloads. Hermes reached this position roughly 90 days after its February 25 launch, accumulating more than 140,000 GitHub stars and nearly 1,000 contributors in that window.
What Triggered the Shift
The lead change followed the May 7 release of Hermes v0.13.0, codenamed “Tenacity,” which shipped 864 commits, 588 merged pull requests, and contributions from 295 developers. Key additions included a Kanban-style durable task board with heartbeat monitoring and hallucination recovery for multi-agent workflows, a /goal command that locks the agent on a target across turns, and Google Chat as the 20th supported messaging platform. The release also closed 8 high-priority security issues.
On the OpenClaw side, the period coincided with ongoing security turbulence. MarkTechPost reported that nine CVEs disclosed in a four-day window in March 2026 (one scoring 9.9 CVSS) and a Koi Security audit finding 341 malicious entries among 2,857 ClawHub skills created an opening for alternatives. The departure of founder Peter Steinberger to OpenAI in February left questions about the project’s direction, though it now operates under an independent foundation with OpenAI as financial sponsor.
Architectural Divergence
The two agents represent opposing bets on what makes an autonomous agent valuable. OpenClaw connects to more than 50 messaging platforms through a central WebSocket gateway and hosts over 44,000 community-built skill files on ClawHub. Its thesis: an agent’s value scales with the breadth of its reach and library.
Hermes supports 20 platforms deliberately and focuses on self-improvement. After every task involving five or more tool calls, the agent runs a reflection step and generates a reusable skill file. An autonomous background process called the Curator grades and rewrites underperforming entries weekly. According to Tech Times, Nous Research’s internal benchmarks show agents with 20 or more self-created skills complete similar future tasks 40% faster than fresh instances, though improvement is domain-specific and does not transfer across unrelated task types.
Both projects use Anthropic’s SKILL.md format, published as an open specification in December 2025 and now adopted by 32 tools including Google’s Gemini CLI, JetBrains Junie, and AWS Kiro.
Daily vs. Cumulative
A single day at the top of OpenRouter does not constitute permanent market leadership. OpenClaw’s cumulative advantage (9.17T vs 6.35T tokens) reflects months of established production deployments that do not migrate overnight. The daily metric signals developer sentiment on new projects and experimental workloads. Whether Hermes can maintain the daily lead through the coming weeks will determine if this is a milestone or a moment.