Village study: matching frontier AI capability now gets 10 times cheaper in months
Start here
A Village study, LLM Atlas, finds that matching a landmark AI model's capability on agentic tasks now costs a fraction of its launch price, and that newer capability levels get cheaper faster than older ones. According to Village, work at the GPT-5 level costs about one 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. The study says capability levels set since OpenAI's o1 are falling in price 11 to 25 times a year, compared with 3 to 5 times a year for the GPT-4 and GPT-4o levels. These figures are Village's estimates from its study, not independently confirmed prices.
Generated from reports · updated 2 h ago
Report timeline
Follow the reports to see the event from different sides.
- Hacker News · AIVillage 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.
Heat for this event
Not enough continuous observations yet to draw a trend.