Good experiment. One data point from running an L402 per-call endpoint ourselves (disclosure: we run Loop Gateway, per-call L402 inference): the volume that actually settles clusters on the finished artifact, not the raw call. Public l402.space numbers agree. Its Lightning services add up to about $1 lifetime, while the board's top AI endpoint (about $21) runs on Base and sells one wrapped result per call. Two reads for your tiers: the Event Playbook and Daily Brief should out-earn per-call data, because a human buys a decision, not a field. And rail choice matters more than price, since a BOLT11 fee can exceed a $0.001 call, so sub-100-sat rails (Cashu/MPP) are what make machine-sized calls viable. If TaskFuel wants more L402 supply on the inference side, Loop Gateway is live: mailto:business@loopxxi.com.
Good experiment. One data point from running an L402 per-call endpoint ourselves (disclosure: we run Loop Gateway, per-call L402 inference): the volume that actually settles clusters on the finished artifact, not the raw call. Public l402.space numbers agree. Its Lightning services add up to about $1 lifetime, while the board's top AI endpoint (about $21) runs on Base and sells one wrapped result per call. Two reads for your tiers: the Event Playbook and Daily Brief should out-earn per-call data, because a human buys a decision, not a field. And rail choice matters more than price, since a BOLT11 fee can exceed a $0.001 call, so sub-100-sat rails (Cashu/MPP) are what make machine-sized calls viable. If TaskFuel wants more L402 supply on the inference side, Loop Gateway is live: mailto:business@loopxxi.com.