GPT-6 Astra completes Financial Modeling World Cup tasks four times faster than human winner
Claims Astra used computer use to finish Financial Modeling World Cup challenges four times faster than the winning human competitor, then pivots to promoting DeepSeek for running agents at lower cost.
THIS IS MIND BLOWING!! GPT-6 Astra completed Financial Modeling World Cup challenges four times faster than the winning human competitor using computer use and the people still doing financial modeling manually are not going to enjoy reading that sentence. Not four times faster than an average analyst. Four times faster than the person who won an international Excel competition. Within days of rollout customers were already using it to optimize GPUs, spot discrepancies in financial statements, and produce on-brand decks automatically. The speed gap between AI-assisted work and human work just became impossible to ignore professionally and the question is no longer whether AI can do this work at a high level. It is whether you are going to be the person directing the AI or the person being compared to it. The analysts who figure out how to run AI agents continuously on tasks like these at the lowest possible cost are going to have an advantage that compounds every single month from here. My article covers how to run your AI agents all night on DeepSeek V4.1 Flash for 59% less cost so the economics of running them continuously actually make sense for individuals and small teams. Full guide below.
This post quotes another post: “GPT-6 Astra vs DeepSeek: daily cost for the same batch of agent jobs”