Author(s): MKWriteshere Originally published on Towards AI. “Why does my Claude chat hit the limit so quickly?” This question echoes across forums as users struggle with the AI assistant’s seemingly arbitrary constraint Claude is the best for creative writing, and it has the worst free tier. Now, a GitHub leak may explain why: Claude’s system prompt is reportedly a staggering 24,000 tokens long, an invisible data colossus that devours your message allowance before you’ve even started your conversation. Even if this leak is fake, the ones on their website are still way too damn long. The official documentation itself includes absurdities like embedding links to prompt engineering guides and paragraphs explaining when to use certain features, content that never surfaces in actual conversations. “Why would a system prompt include that?” Imagine an Olympic ice skater gliding effortlessly across the ice. The audience sees only the graceful performance, completely unaware of the rulebook constraining every movement. Similarly, when chatting with Claude, you experience the polished surface of an AI whose every response is governed by a bloated instruction set so extensive that users hit “your chat is getting too long” warnings after just two messages on the free tier. Today, we’re dissecting this leaked behemoth, examining the excessive instructions that determine what Claude… Read the full blog for free on Medium. Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor. Published via Towards AI
↧
