Latest open artifacts (#25): DeepSeek V4.1 Flash 1017, American-made models, and new regulatory controls in the US & China
DeepSeek released internal inference stack DeepInfer as US and China both impose new AI regulatory controls.
“we consider this release to be one of the bigger open-source AI artifacts in the last few months”— Nathan Lambert
DeepSeek publicly released DeepInfer, its internal inference stack — the first time a leading AI lab has shared such tooling — with estimates pointing to ~90% gross margins on token sales, which will pressure pricing across the ecosystem. Simultaneously, the US DHS banned DeepSeek from federal civilian procurement, and China's MOFCOM proposed a pre-approval framework for international sharing of frontier AI models. The convergence of inference efficiency disclosure and accelerating bilateral regulatory controls marks a pivotal inflection point for the open-weight AI geopolitical landscape.