Anthropic Releases Claude 4 with Extended Context and Autonomous...

Anthropic released Claude 4, its most capable AI model family to date, featuring a one-million-token context window that can process entire legal libraries, codebases, or multi-year research corpora in a single prompt. The model introduces autonomous agent capabilities, allowing it to browse the web, write and execute code, manage files, and coordinate multi-step workflows without human intervention between steps. Claude 4 scored state-of-the-art results on the MMLU Pro, GPQA Diamond, and HumanEval benchmarks, surpassing all previously published models on graduate-level scientific reasoning tasks. Anthropic also released the Claude 4 API to enterprise customers in India and Southeast Asia as part of a dedicated emerging markets programme at subsidised token rates. The company said Claude 4 embeds Constitutional AI principles at a deeper architectural level than any previous model, making refusals more contextually nuanced and factual outputs more reliable.