Anthropic Launches Claude 4 with Enhanced AI Capabilities
Anthropic unveiled its latest AI models, Claude Opus 4 and Claude Sonnet 4, marking a significant advancement in artificial intelligence capabilities. Claude Opus 4, described as Anthropic's most powerful model to date, excels in coding and complex problem-solving tasks. It achieved a 72.5% score on the SWE-bench coding benchmark, surpassing competitors like OpenAI's GPT-4.1 and Google's Gemini 2.5 Pro. The model is capable of maintaining focused effort across extended tasks, operating autonomously for up to seven hours, which significantly expands the potential for AI agents in various applications .
Claude Sonnet 4, while more affordable, offers substantial improvements over its predecessor, Claude Sonnet 3.7. It delivers enhanced coding and reasoning abilities, making it suitable for general tasks and high-volume operations. Both models introduce "thinking summaries," a feature that condenses the AI's reasoning process into understandable insights, and an "extended thinking" mode that allows users to toggle between rapid responses and deeper analytical reasoning .
In response to the advanced capabilities of Claude Opus 4, Anthropic has implemented heightened safety measures. The model is the first to trigger the company's Responsible Scaling Policy at AI Safety Level 3 (ASL-3), due to concerns about potential misuse in areas like bioweapons development. These safeguards include enhanced cybersecurity protocols, anti-jailbreak measures, and a bounty program for identifying vulnerabilities. Anthropic's proactive approach aims to set a precedent for responsible AI deployment in the industry .
