Editorial illustration for Hermes Agent tops use as Nous Research’s self‑improving model leads OpenRouter
Hermes Agent tops use as Nous Research’s self‑improving...
Hermes Agent tops use as Nous Research’s self‑improving model leads OpenRouter
Why does this matter? As of May 10, 2026, Hermes Agent—an open‑source model from Nous Research—has claimed the top spot on OpenRouter’s global daily app and agent rankings. The shift isn’t just a number; the agent now processes 224 billion tokens a day, edging out OpenClaw’s 186 billion.
Here’s the thing: the two projects embody opposite design philosophies. While OpenClaw relies on a central WebSocket Gateway that stitches together more than 50 messaging channels—Telegram, Discord, Slack, WhatsApp, Signal and others—its focus is sheer reach across platforms. Hermes, by contrast, runs under an MIT license and follows a “do, learn, improve” loop.
After each task it pauses, reflects and auto‑generates reusable skill files. Memory is split into three layers: a persistent user‑agent snapshot, a SQLite FTS5 full‑text search of every session, and procedural skill files that capture repeatable logic. Adding to the narrative, OpenClaw’s founder Peter Steinberger joined OpenAI in February 2026, and the project now sits under an independent open‑source foundation sponsored by the same company.
The rivalry, therefore, is less about popularity contests and more about where developers see the future of autonomous agents.
Why this matters
Hermes Agent now sits at the top of OpenRouter’s daily app and agent rankings, pulling 224 billion tokens a day compared with OpenClaw’s 186 billion. The shift suggests a sizable slice of the developer community is currently favoring the depth that Nous Research’s self‑improving model offers. For us, the metric is a concrete signal of usage, not a guarantee of superiority; daily token volume can fluctuate with project cycles or promotional bursts.
Still, the overtaking of a previously dominant open‑source agent raises questions about where engineering effort will concentrate next. Developers may see Hermes as a reference point for building richer interactions, while founders might weigh its traction against the maturity of alternatives. Researchers should note that the “meaningful portion” of the community betting on depth is, for now, a snapshot—whether this translates into broader overall impact remains unclear.
We’ll keep watching the rankings for signs of sustained momentum.
Further Reading
Related Reading
Research & Benchmarks DeepMind spinoff’s AI‑designed drugs enter human trials after AlphaFold 3
Research & Benchmarks Hyperparameter Tuning Reaches 0.9617 Accuracy in 64.59 Seconds
Research & Benchmarks Google AI Advisors Let Users Probe Performance with Conversational “Why” Queries
From Archives Palisade Research: Open‑weight AI like Qwen boost autonomous hacking
From Archives