Anthropic expanded Claude's context window from 9,000 to 100,000 tokens — roughly 75,000 words — and made it available through its API. Hundreds of pages of material, or an entire book, could now fit in a single prompt, enabling use cases such as analyzing financial statements, reviewing legal documents, and prototyping against a whole codebase. In a demonstration, Claude read the full text of The Great Gatsby and identified a modified line in 22 seconds. The announcement staked out long-context handling as a differentiator, and context length became a standard axis of model comparison thereafter.