Anthropic has released Sonnet 4.6, the latest iteration of its Claude AI model, as the company faces mounting pressure from Pentagon officials over its AI safety and deployment policies. The release comes amid reports of tensions between the defense department and Anthropic regarding the company's stance on military applications of artificial intelligence. Anthropic CEO Dario Amodei has publicly stated that threats from the Pentagon will not sway the company's position on responsible AI development and deployment. Concurrently, Google has rolled out Gemini 3.1 Pro, intensifying competition in the large language model space. The competing releases highlight the rapidly evolving landscape of advanced AI models, with major tech companies racing to deliver increasingly capable systems. Anthropic's introduction of deep-thinking tokens in its latest offering represents a technical advancement aimed at improving model reasoning capabilities. The standoff between Anthropic and the Pentagon underscores broader tensions in the tech industry regarding government contracts, military AI applications, and corporate values. As defense agencies seek to leverage cutting-edge AI capabilities for national security purposes, some AI developers are establishing ethical boundaries around their technology's potential military uses.