Get the latest tech news
Andrew Ng on DeepSeek
The Batch AI News and Insights: The buzz over DeepSeek this week crystallized, for many people, a few important trends that have been happening in...
Driven in part by the U.S. AI chip embargo, the DeepSeek team had to innovate on many optimizations to run on less-capable H800 GPUs rather than H100s, leading ultimately to a model trained (omitting research costs) for under $6M of compute. On OSWorld, a benchmark that evaluates the ability of multimodal agents to perform complex tasks that involve real-world web and desktop apps, CUA achieved a success rate of 38.1 percent. The order directs the federal Office of Management and Budget to award government contracts to AI companies that align with the administration’s emphasis on advancing U.S. competitiveness and national security.
Or read this on Hacker News
