Claude Fable now requires usage credits

**Claude Fable** now requires **usage credits**, according to a report shared on X by **JJordan** and picked up on Hacker News under a Claude/Codex/Fable thread. The HN item sits at 1 point with 1 comment, so the signal is early and thin, but the core claim is clear: access that previously did not gate on credits now does.

There is no public technical breakdown in the given sources of how the credit ledger is enforced, how rates map to tokens or sessions, or whether the change is client-side, account-tier, or API-side. What builders can treat as solid is the product mechanic itself: **usage credits** become a required unit of spend for Fable, which turns free or unlimited trial-style use into metered consumption.

For engineers and builders, that matters because any workflow that treated Fable as an always-on sandbox—prototyping agents, long coding sessions, or side-by-side comparison with **Claude** or **Codex**—now has a hard resource bound. Credit exhaustion becomes an operational failure mode: jobs stop mid-run, eval loops need budgets, and cost must be planned before you scale parallel runs.

On the competitive side, the framing on HN already groups **Claude**, **Codex**, and **Fable** together. Metered credits are the common pattern across coding and agent tools: vendors shift from open access to usage-linked pricing as demand and model cost rise. Fable joining that model narrows the gap between “try freely” and “pay per use,” and it forces comparisons on effective cost per useful session, not just model quality.

Practical takeaway: treat Fable as a budgeted resource from here. Instrument usage, set per-project credit caps, and decide which tasks stay on Fable versus Claude or Codex based on credit burn and output quality. Watch next for how credits are priced, whether free tiers or grants remain, and whether the HN discussion grows past this single-point, single-comment notice into confirmed policy or docs from the product side.