Anthropic's flagship model, Fable 5, resumed global access in early July after nearly three weeks of being blocked by US export controls. The model excels in long-duration, multi-step agent tasks, effectively reducing goal drift, but its overly stringent safety filtering mechanisms have sparked controversy. Testing shows that in fields such as biomedicine and AI model development, many legitimate academic research requests are frequently intercepted, with the model sometimes refusing to answer even basic scientific questions. Additionally, Fable 5's high cost and relatively slow response times are major drawbacks. Meanwhile, Anthropic also released the cost-effective Sonnet 5 and the research-oriented Claude Science, attempting to support its nearly trillion-dollar IPO valuation through a full-stack product lineup. Fable 5 was relaunched on Claude services in overseas markets such as South Korea on the morning of July 2. The US government had, on June 12, urgently suspended access for foreign nationals to Fable 5 and its sibling model Mythos 5, citing that the models posed "very real risks" to cybersecurity, financial systems, and critical infrastructure. With the export controls lifted on June 30, Anthropic was able to once again offer this cutting-edge model to global users.