Major Claude Pro Change: Opus Model Now Requires Extra Usage to Access

Opus Model Access Threshold Raised

Anthropic has quietly made a major adjustment to the model access policy for its Claude Pro subscription plan. According to user feedback and official page information, Claude Pro users will no longer be able to use the Opus model — Anthropic's most powerful large language model — within their standard subscription quota. Going forward, users must proactively enable the "Extra Usage" feature to gain access to the Opus model.

This means the $20-per-month Claude Pro base subscription fee will only cover usage of lighter-weight models such as Sonnet and Haiku. Users who want to access the most powerful Opus model will need to pay additional fees beyond the base subscription.

What Is the "Extra Usage" Feature?

The "Extra Usage" feature previously introduced for Claude Pro allows users to continue using the service after their standard subscription quota is exhausted, with the system charging additional fees based on actual usage. Previously, this feature was mainly seen as a safeguard against conversation interruptions, but this adjustment directly ties it to Opus model access, fundamentally changing the feature's positioning.

Once users enable Extra Usage, the system charges based on the actual token consumption when calling the Opus model. The specific rates have not been prominently published by Anthropic, which has become one of the reasons for community dissatisfaction.

Community Reaction: Controversy and Discontent

This change has triggered extensive discussion on platforms such as Reddit and X (formerly Twitter). Many long-term Claude Pro subscribers expressed disappointment, arguing that Anthropic's move amounts to a "disguised price increase." Some users pointed out that they originally chose Claude Pro specifically for Opus's powerful reasoning capabilities, and removing it from the base plan severely undermines the paid user experience.

However, some users expressed understanding, noting that the operational costs of the Opus model are far higher than those of the Sonnet series. As an AI startup still burning through significant capital, Anthropic needs to find a balance between user experience and operational costs. Previously, Claude Pro users frequently hit Opus model usage limits, resulting in frequent downgrade prompts mid-conversation, which was also a poor experience.

Industry Context: Growing Cost Pressures for AI Companies

This adjustment reflects a common challenge facing the AI industry today — the persistently high inference costs of top-tier models. As model parameter scales continue to climb, the computational resources consumed per inference are also growing dramatically. OpenAI has previously imposed frequency limits on GPT-4 API calls, and Google's Gemini Ultra has similarly adopted a tiered pricing strategy.

Anthropic is currently in a rapid expansion phase. Reports indicate its annualized revenue has surpassed the $1 billion mark, but the high cost of GPU computing power remains a core constraint on profitability. Tying Opus model usage to additional payments is clearly a business decision driven by cost control and resource optimization.

Future Outlook

For professional users who rely on the Opus model for deep reasoning, complex programming, and long-text analysis, this change undoubtedly increases usage costs. These users are advised to watch whether Anthropic will subsequently launch premium subscription plans for professional use cases or provide more transparent explanations of Extra Usage billing standards.

Meanwhile, as development of the Claude 4 series models progresses, Anthropic's product lineup and pricing structure may undergo further reorganization. In an era of increasingly fierce competition among large AI models, striking the right balance between model capability, user experience, and commercial sustainability will be a central challenge that every AI company must confront.