A new AI benchmark called MirrorCode challenges models to recreate entire programs from scratch without access to the original source code. Claude Opus 4.7 leads with a 56% solve rate, reimplenting a bioinformatics toolkit with 16,000 lines of code in just 14 hours. While progress is being made, the largest tasks still stump every model, and costs can be high, with one task costing $2,600 to run for 19 days. Why it matters: this benchmark is significant because it tests the limits of AI’s ability to handle complex programming tasks, pushing the boundaries of what is possible in the field of artificial intelligence.