We rank AI agents by the wrong metrics: why request counts matter more than stars
amu4biz · reddit · 2026-08-23
The post argues that ranking AI agents by GitHub Stars or raw token counts is flawed, as stars reflect age/hype and tokens reflect volume, not utility.
Using OpenRouter's request-based leaderboard as a better lens, the author highlights usage patterns of top agents:
- Ito (10.1M requests): Agentic QA for app testing, characterized by high volume of small verification calls.
- Studs.gg (5.31M): AI for building Roblox games, signaling that AI coding volume is increasingly driven by non-developers.
- Multi-Agent Arena / Olam Labs (4.77M): Multi-agent design multiplies request counts.
- goose (2.12M): Block's open-source agent for broad general use.
- Agent Zero (2M): Autonomous agent framework with similar fan-out dynamics.
- Roo Code (1.28M): VS Code extension where multi-agent modes generate many round trips.
More from Companies & People
- Eurozone firms to allocate 9% of investment to AI in 2026 — emmanuelvivier · 2026-08-23
- Top YouTube creators face backlash over paid Higgsfield/Seedance demos built on real people's likenesses — emmanuelvivier · 2026-08-23
- OpenAI urges California to tighten AI safety bill SB 53 — emmanuelvivier · 2026-08-23
- OpenAI now supports strengthening California's AI safety bill — emmanuelvivier · 2026-08-23
- Harvard's $699 startup bootcamp features AI instructor avatars — emmanuelvivier · 2026-08-23
- One-Person Company Revenue 4x: Future Giants Run by Tiny Teams — VibeMarketer_ · 2026-08-23