AI Pulse · Business Models & Winners

Anthropic planned to move Claude Fable 5 from included usage to metered credits on July 7. Then it pushed the date to July 12. Then to July 19. Then it split the change by tier instead of doing it all at once.

Max and Team Premium subscribers keep Fable 5 included, capped at 50% of their weekly limits. Pro and Team Standard users now pay metered credits, $10 or $50 per million tokens depending on tier. Anthropic softened the blow with a one-time $100 credit, but the underlying shift is clear: the most capable model is no longer bundled for most plans.

Why this matters to you

This is a live case study in how hard it is to move customers from seat-based pricing to usage-based pricing, even when you’re Anthropic. Two delays and a tier split show the internal tension between revenue goals and customer backlash. If a well-funded AI lab with a narrow product line struggles to execute this cleanly, expect the same friction when you roll out usage pricing for your own AI features or internal tools.

The pattern is worth watching. Bundled usage feels simple until your best model costs too much to give away. Metered credits feel fair until users complain they can’t predict their bill. Anthropic tried to thread the needle with tier splits and free credits, but the delays signal they didn’t get the messaging or the economics right the first time. When you build AI workflows into your business, whether through an AI command centre like Omni or standalone tools, plan for pricing changes like this. Lock in contracts where you can, budget for overage, and keep fallback models in your stack so one vendor’s pricing pivot doesn’t stop your operations. Read the full breakdown for the tier-by-tier details.