ResearchThe Decoder· Jun 26, 2026
AI Model Runs 19 Days on Single MirrorCode Task Costing $2,600
Epoch AI released the MirrorCode benchmark to evaluate how well AI systems can reconstruct full programs from scratch. Claude Opus 4.7 achieved the highest score at 56 percent and completed one 16,000-line task in 14 hours, yet all tested models failed on the hardest cases. One run extended to 19 consecutive days and incurred a $2,600 compute cost.
Key points
- →MirrorCode measures AI ability to rebuild programs without source access
- →Claude Opus 4.7 leads with 56 percent solve rate
- →Longest single task required 19 days and $2,600 in compute
- →Every model failed the most complex benchmark items
Mentioned
Epoch AIClaude Opus 4.7MirrorCode
Related stories
Google Research Speeds Up Gemini Nano on Pixel DevicesGoogle Research · Research→Former Databricks AI Chief Claims 1,000x Cut in AI Power UseTechCrunch · Research→Token Prediction Analysis for Hybrid ModelsHugging Face · Research→Investigation Finds Left-Leaning Responses in Most Major AI ChatbotsThe Decoder · Research→Authors Guild Tests Show Mixed Results for AI Detectors on Human TextThe Decoder · Research→Google Losing Top AI Researchers to Rival CompaniesThe Decoder · Research→
This is an original summary by Dhanasvi's agents based on The Decoder's public feed. For the complete article, visit the original source. Trademarks and article copyright belong to their owners.