The competition in the AI coding sector is intensifying, with a focus on “context windows,” which represent an AI model’s capacity to retain information while generating responses. In a major advancement, the AI startup Anthropic has announced a significant enhancement to its context window, increasing it fivefold as it aims to keep pace with giants like OpenAI and Google.
Measured in tokens, Anthropic’s Claude Sonnet 4 now features a context window capable of processing 1 million tokens. In comparison, the company’s previous context window of 500,000 tokens could manage around 100 half-hour sales discussions or 15 financial reports, meaning the new window will allow extensive analysis, such as reviewing dozens of research papers or hundreds of documents in one API query, as stated by Anthropic.
Most notably, the coding capabilities of the AI have vastly improved, enabling the handling of entire codebases ranging from 75,000 to 110,000 lines of code, a significant leap from the previous threshold of 20,000 lines.
Brad Abrams, the product lead for Claude, emphasized the significance of this upgrade in an interview with Technology News. He remarked, “This is exciting because it addresses a major obstacle faced by our customers. Previously, they had to break their projects into smaller segments due to our prior context limitations. The expanded 1 million token capacity allows the model to grasp the entire context and tackle challenges on a larger scale.”
Abrams further noted that Sonnet 4 can now accommodate approximately 2,500 pages of text, even capable of encompassing a full edition of Tolstoy’s War and Peace.
However, Anthropic is not alone in offering such a vast context window; OpenAI’s recent release of GPT-4.1 has set a similar benchmark.
For companies like Anthropic and OpenAI, enterprise clients are proving to be a lucrative source of income—especially as they compete in the lucrative coding assistance market. Both companies have been in an ongoing race to develop and launch new features while attempting to outmaneuver one another. Recently, OpenAI introduced GPT-5, highlighting its coding efficiency against competitors. In this context, Anthropic seeks to regain its competitive edge, as it reportedly pursues funding that could increase its valuation to as high as $170 billion.
Abrams indicated that clients from various fields, including coding, pharmaceuticals, retail, professional services, and legal sectors, have expressed strong interest in the newly enhanced context window.
When asked whether the timely release of OpenAI’s GPT-5 influenced Anthropic’s decision to expedite its context window launch, Abrams responded, “We’re moving quickly and responding to customer feedback. Just two and a half months ago, we unveiled Opus 4 and Sonnet 4, and last week we rolled out Opus 4.1. Now, with this million token context, we’re showcasing our commitment to delivering these enhancements as rapidly as possible.”
The updated context window is currently available through the Anthropic API for select customers, particularly those with Tier 4 status and custom rate limits, indicating a significant investment in the platform. Wider access is expected to be introduced in the forthcoming weeks, according to the company’s blog announcement.
