Village study: matching frontier AI capability now gets 10 times cheaper in months

Village's LLM Atlas study finds that matching a landmark model's capability on agentic tasks now costs a fraction of its launch price, and newer levels fall faster. GPT-5-level work costs a thirty-third of its launch price, and an agentic task that cost 79 cents with GPT-5 in August 2025 costs about 2 cents with OpenAI's GPT-6 Luna. Levels set since o1 are falling 11 to 25 times a year, versus 3 to 5 times a year for GPT-4 and GPT-4o levels.