Anthropic tests AI model more powerful than Mythos, says no release plans yet

Anthropic, in its latest risk report, said it is internally testing an AI model called "Model 2", which is "somewhat more capable than" its most powerful model, Mythos 5. However, the company said it has no current plans to release it. "Model 2 is used heavily within Anthropic for coding, data generation, and other agentic use cases," the report said.