Confidential networks replaced with domestic alternatives? Palantir, Cohere, Mistral probably laughing secretly.

According to a report from the Coin World network, the Pentagon is testing multiple artificial intelligence models to determine which models its 25 “super users” prefer, while accelerating the search for a replacement for Anthropic’s Claude. Testing began in early March—just three days after Defense Secretary Hegseth announced Anthropic as a supply-chain risk. After parting ways with Anthropic, the Pentagon gave itself six months to gradually phase out the company’s products. Earlier this month, the Pentagon announced new agreements with several other companies to use their AI tools on classified networks, seeking to develop multiple new vendor models.