73% of every API call is fixed overhead (I measured it)
I built a monitoring dashboard to profile token consumption on a Hermes v0.6.0 deployment running Telegram + WhatsApp + Cron gateways. After analyzing 6 request dumps, I found that 73% of every API call is fixed overhead.
— @Bichev, 2026, via GitHub.