Anthropic Releases Sonnet 4.6 AI Model Featuring Major Context Window Expansion

According to the Economic Desk of Webangah News Agency, Anthropic announced the debut of its latest artificial intelligence model, Sonnet 4.6, highlighting key enhancements across several performance metrics. In its official introduction, the company stressed notable upgrades in the model’s capacity for writing code, adhering to complex instructions, and utilizing computer tools.

The Sonnet 4.6 model is designated as the standard offering for individuals utilizing both the Free and Pro subscription tiers. This release follows closely on the heels of the Opus 4.6 model launch just two weeks prior, with expectations high for the subsequent unveiling of an updated Haiku model in the coming weeks.

A central feature of the Sonnet 4.6 beta version is the introduction of a substantially larger context window, now featuring one million tokens. Anthropic noted that this capacity is double the largest context window previously available for Sonnet. The firm characterized this expanded window as capable of accommodating extensive codebases, lengthy legal contracts, or dozens of research papers within a single processing request.

The new model is entering the market backed by a suite of new benchmark scores, including record-breaking results in computer usage benchmarks such as OS World and SWE-Bench for software engineering tasks. Perhaps the most impressive metric is the model’s score of 60.4 percent on the ARC-AGI-2 benchmark, a test specifically designed to measure human-like reasoning skills. This performance places Sonnet 4.6 ahead of most directly comparable models, although it continues to trail behind leaders like Opus 4.6, Gemini 3 Deep Think, and a refined iteration of GPT 5.2.