{"version":1,"type":"story","url":"https://digestai.news/story/ai-performance-costs-have-dropped-about-47-per-quarter-report-finds","json":"https://digestai.news/story/ai-performance-costs-have-dropped-about-47-per-quarter-report-finds.json","markdown":"https://digestai.news/story/ai-performance-costs-have-dropped-about-47-per-quarter-report-finds.md","slug":"ai-performance-costs-have-dropped-about-47-per-quarter-report-finds","headline":"AI performance costs have dropped about 47% per quarter, report finds","summary":"A new Epoch AI report by Emberson and Roodman says the cost of achieving a given level of AI performance has fallen an average of some 47% each quarter over the past three years, a 13‑fold drop per year. The authors compare OpenAI’s o3 model, which cost about $0.30 per question to reach 75% on GPQA Diamond in January 2025, with GPT‑5.6 Luna, which achieved the same score for roughly $0.0004 in mid‑2026 – a roughly 725‑fold decline in under 18 months.\n\nThe report argues that cheaper inference at a fixed performance level reduces the threat from open‑source models, because frontier models become not only smarter but also dramatically cheaper to run. This rapid cost decline is presented as faster than any other transformative technology in history.","keyPoints":["Cost per performance level fell about 47% each quarter, a 13‑fold yearly drop, per the report.","OpenAI o3 cost $0.30/question for 75% GPQA Diamond in Jan 2025; GPT‑5.6 Luna cost $0.0004 in mid‑2026.","The 725‑fold cost reduction in under 18 months suggests cheaper inference for frontier models."],"whyItMatters":"Rapidly falling AI inference costs could lower barriers for developers and reduce competitive pressure from open‑source models, reshaping deployment economics across the industry.","category":{"slug":"models","name":"Generative AI & Models","url":"https://digestai.news/category/models"},"entities":{"companies":["OpenAI"],"models":["o3","GPT-5.6 Luna"],"people":["Emberson","Roodman"]},"firstPublishedAt":"2026-09-23T13:08:59Z","updatedAt":"2026-09-26T06:38:00Z","sourceCount":2,"hasPrimarySource":false,"sources":[{"outlet":"marginalrevolution.com","title":"The Price of Intelligence Is Falling Rapidly","url":"https://marginalrevolution.com/marginalrevolution/2026/09/the-price-of-intelligence-is-falling.html","publishedAt":"2026-09-23T13:08:59Z","type":"press","primary":false,"lead":true},{"outlet":"note.com","title":"Before Trying GPT-6 Astra, Compare It with Your Own Work | 3 Tasks with Answer Keys and an Adoption Decision Sheet","url":"https://note.com/mio_change_26/n/n7d25e0dedc57?hl=en","publishedAt":"2026-09-26T06:38:00Z","type":"press","primary":false,"lead":false}],"sourceNotes":null,"discussions":[{"site":"Hacker News","url":"https://news.ycombinator.com/item?id=49815526","points":75}],"thread":{"title":"OpenAI Unveils GPT-6 Sol and Luna Models","url":"https://digestai.news/thread/comparing-openai-models-on-aws-for-workload-efficiency","storyCount":6},"cite":{"text":"Digest AI, \"AI performance costs have dropped about 47% per quarter, report finds\", 23 September 2026, https://digestai.news/story/ai-performance-costs-have-dropped-about-47-per-quarter-report-finds","publisher":"Digest AI","title":"AI performance costs have dropped about 47% per quarter, report finds","datePublished":"2026-09-23T13:08:59Z","url":"https://digestai.news/story/ai-performance-costs-have-dropped-about-47-per-quarter-report-finds"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}