Anthropic raised Claude Code's weekly usage limit by 50 percent on May 13, but refused to disclose what that limit actually is in concrete terms. The company has never published the token size of any Claude Code limit, maintaining a pattern of exact pricing transparency paired with deliberate opacity around usage caps. This strategy reveals how AI companies are managing the gap between compute costs and subscription revenue as demand scales.

Why Is Anthropic Hiding the Numbers Behind Claude Code Limits?

Anthropic publishes precise per-token API rates for Claude Opus 4.7, Sonnet 4.6, and Haiku 4.5 models. It also lists exact monthly subscription prices: $20 for Pro, $100 for Max 5x, and $200 for Max 20x. Yet when it comes to Claude Code usage caps, the company offers only vague references. The widely cited figure of 40 to 80 hours of Sonnet per week traces back to a July 2025 statement that has never been updated, even though the actual limits behind it have shifted repeatedly.

The May 13 increase followed a May 6 announcement that doubled Claude Code's five-hour session limits and removed weekday peak-hours throttling, paired with news that Anthropic had leased SpaceX's Colossus 1 data center in Memphis. The timing suggests the subscription limit and compute lease were coordinated announcements, hinting that new capacity enabled the expansion.

How Does Anthropic's Pricing Model Actually Work?

• Transparent API Costs: Anthropic publishes exact per-token rates for all Claude models, allowing developers to calculate costs precisely for API usage.
• Opaque Subscription Caps: Monthly subscription tiers are priced as multipliers of the Pro plan, but the token allowances behind each tier are never disclosed publicly.
• Dynamic Limit Adjustments: Anthropic re-tunes usage caps as compute capacity arrives or runs out, communicating changes through announcements rather than published specifications.

The economics underneath this approach are straightforward: a $20 Pro subscriber running Claude Code intensively costs Anthropic close to $60 in compute resources. Vague limits become the lever Anthropic adjusts as infrastructure expands or contracts. By not publishing token counts, the company avoids committing to specific guarantees while maintaining the appearance of generosity when it raises caps.

July 13, 2026, marks the next checkpoint. Subscribers will learn whether the 50 percent increase becomes permanent or lapses the same way they learned of every previous change: through an announcement with no published denominator. This pattern suggests Anthropic views usage limits as a dynamic tool for managing compute demand rather than a fixed contractual promise.

The strategy contrasts sharply with how the company handles pricing. Anthropic's willingness to publish exact API rates while withholding usage cap details signals that token economics are a competitive advantage worth protecting, while subscription pricing is a marketing tool worth standardizing. As Claude's adoption grows and compute costs remain high, this opacity may become a friction point for enterprise customers seeking predictable budgeting.