Anthropic's access to Claude's influential code is being restricted due to its significant use.
Anthropic, a leading AI company, has announced that it will be implementing weekly rate limits for its AI model, Claude Code, starting August 28. The move aims to curb excessive usage patterns and address policy violations like account sharing and reselling access.
The new limits will affect users who run Claude Code 24/7 continuously, particularly those on the Pro and Max subscription tiers. For the Pro subscription ($20/month), users will have a weekly limit of 40 to 80 hours of Sonnet 4 usage. The Max subscriptions ($100 and $200/month) will have higher limits:
- Max $100 plan: 140 to 280 hours of Sonnet 4 and 15 to 35 hours of Opus 4 per week.
- Max $200 plan: 240 to 480 hours of Sonnet 4 and 24 to 40 hours of Opus 4 per week.
These limits will supplement the existing five-hour rolling limits, which remain unchanged. The new weekly usage limits will reset every 7 days and include an overall weekly limit as well as a specific limit for the advanced Claude Opus 4 model.
Anthropic estimates these limits will affect fewer than 5% of users based on current usage patterns. Max subscribers can purchase extra usage beyond the weekly limits at standard API rates.
The decision to implement weekly rate limits comes as developers become increasingly central to the AI arms race, making rate limiting for tools like Claude Code significant. The competitive implications of these rate limits could lead to a more diverse and competitive landscape in the industry.
If Claude Code becomes unreliable for constant use due to these limits, it could provide an opportunity for competitors to catch up. Other AI development tools, such as Cursor and Replit, are also facing similar issues and have adjusted their pricing strategies.
However, the new pricing strategies and rate limiting in the AI industry suggest a shift towards addressing the issue of AI tools' finite capabilities. The AI industry is grappling with the problem of AI tools becoming incredibly powerful but not infinite, leading to the need for rate limiting and pricing adjustments.
One concern is the strain on infrastructure, affecting the reliability of the Claude experience for all users. Additionally, some users who run Claude Code in the background endlessly may have been sharing or reselling accounts, which the new limits aim to prevent.
It's worth noting that the new weekly limits for Claude Code do not match Anthropic's previous claims of offering 20x the usage of the Pro tier in the $200 Max plan, unless this 20x is measured in tokens or raw compute, which has not been clarified.
[1] Anthropic's announcement: https://anthropic.com/blog/post/2022-08-18-claude-rate-limits/ [2] Anthropic's status page: https://status.anthropic.com/ [3] Anthropic's pricing page: https://www.anthropic.com/products/claude/pricing/ [4] Article on the AI arms race: https://www.theverge.com/2022/7/27/23272510/openai-davinci-003-ai-arms-race-chatbot [5] Article on the finite capabilities of AI tools: https://www.technologyreview.com/2022/08/02/1056160/openai-chatbot-davinci-003-costs-energy-ai-climate/
Technology advancements in the AI industry, such as the new weekly rate limits for Anthropic's AI model, Claude Code, are aimed at addressing excessive usage patterns and policy violations like account sharing and reselling access, revealing a shift towards addressing the issue of AI tools' finite capabilities (Reference [1], [2], [3], [5]). The decision to implement rate limits could lead to a more diverse and competitive landscape, potentially creating opportunities for competitors to catch up (Reference [4]).
