Anthropic’s 50% weekly usage limit increase for Claude Code — covering Pro, Max, Team, and seat-based Enterprise plans, across CLI, IDE extensions, desktop, and web — runs out at 6pm PDT on 13 July 2026. It followed a doubling of the shorter 5-hour usage window the week before, making it the third capacity intervention inside roughly five weeks. Anthropic hasn’t said what the ceiling reverts to afterward.
Why a model company is fighting on quota, not just capability
For most of 2025, competition between AI coding tools was framed as a benchmark race — whose model wrote better code. That’s flattened out: leading models now cluster within a narrow band on standard coding benchmarks. With capability differences narrowing, the competitive lever that’s actually moving is how much of a given tool you can use before hitting a wall, and at what price. Anthropic’s repeated, stacked capacity increases are a direct response to OpenAI’s Codex gaining ground on usage terms, not benchmark wins.
Why this matters if you’re commissioning a build
If your development partner’s AI-assisted workflow is priced or scoped around current usage limits, a reversion in mid-July is worth asking about upfront — not because the sky is falling, but because “AI coding tool economics” is currently a fast-moving target, and a team that hasn’t priced in the volatility is guessing at margins rather than managing them. This is also a reminder that raw usage capacity, not just model quality, is now a genuine variable in how much agentic coding work a team can responsibly commit to in a sprint.
So what
We treat AI tool quotas and pricing as a live input to project scoping, not a fixed cost we absorb quietly — reviewed alongside which models and context sizes a given piece of work actually needs. More on how that fits our delivery process on the AI-assisted development page, or get in touch if you want a straight answer on what a project should cost.