OpenAI and Anthropic double Codex and Claude limits for holidays

Holiday capacity boosts for Codex and Claude might sound like a marketing gimmick, but they reveal something important about the current bottlenecks in the AI stack. Both OpenAI and Anthropic are signaling that, under normal conditions, they run these systems close enough to their infrastructure ceiling that doubling limits is notable and time‑bounded. Their willingness to temporarily eat more compute for enthusiasts hints at both confidence in their scaling plans and sensitivity to user frustration with tight quota walls.

From an ecosystem perspective, higher short‑term limits create a burst of experimentation. More developers can stress‑test agentic coding workflows, CI integrations and long‑running dev sessions without immediately bouncing off rate caps. That generates data about real‑world usage patterns, which in turn can inform pricing, product segmentation and the design of future “Pro” or “Max” tiers.

This kind of demand shaping is likely to be a recurring feature as labs try to ration frontier model access while they race to build the next generation. It underlines that, even as models become more capable, compute scarcity and access policies remain key levers in who gets to push the frontier forward.