Anthropic is preparing to launch Claude Opus 4.7 and an AI-powered design tool for building websites and presentations, a move that rattled shares of Adobe, Wix and Figma on Monday.
Opus 4.7 Launch Details
The Information reported on Apr. 14 that both products could ship as early as this week, citing a person with knowledge of the plans. The design tool accepts plain English prompts and targets developers and non-technical users alike.
That puts Anthropic in direct competition with startups like Gamma and Google's Stitch.
Opus 4.7 is not even Anthropic's most advanced model.
That distinction belongs to Claude Mythos, a cybersecurity-focused system the company is distributing only to select security firms through Project Glasswing while keeping it away from the general public.
Also Read: World Liberty Financial Demands Insiders Burn 10% Of Their WLFI Or Stay Locked
AI Benchmark Crisis And Mythos
OpenAI recently declared SWE-bench Verified, the leading coding benchmark, "contaminated" after finding that frontier models had memorized its solutions during training. Yet labs continue citing those same tests in model comparisons.
A separate ARC-AGI-3 evaluation underscored the gap between AI systems and human reasoning. Gemini scored 0.37% and GPT-5.4 hit 0.26%, while humans achieved 100%. Without a detailed model card from Anthropic, claims about Opus 4.7's improvements remain difficult to verify independently.
The UK's AI Security Institute recently evaluated Mythos Preview and found it can autonomously execute sophisticated cyberattacks at rates no other model has matched. It became the first AI to complete "The Last Ones," a 32-step corporate network attack simulation that typically takes human red teams 20 hours.
Read Next: Binance Launches Built-In Chat Feature To Merge Messaging With Crypto Transfers
