Inference Prices Plunge 214x in Two Years: The Collapse of Intelligence Cost
McDonaghMatthew · x · 2026-08-08
The author argues that we have been inside the technological singularity since 2022, highlighted by the dramatic collapse in the cost of intelligence. Historically bound to human labor, acquiring more analysis or engineering required hiring more people—a paradigm now breaking down.
According to data from Epoch AI, the lowest API price to achieve GPT-4-level performance (on the MMLU benchmark) plummeted from $37.50 per million weighted tokens in March 2023 to just $0.175 in February 2025.
This marks a 214x cost reduction for comparable intelligence in under two years. The article emphasizes this isn't just a single model getting cheaper, but the market learning to deliver capable intelligence at a fraction of the cost. Once inference becomes inexpensive enough to stop rationing, it will be embedded across every product, workflow, and decision.
Related event: Cost of GPT-4-Level AI Plummets 99.5% in Two Years(2 posts)→
More from AGI Musings
- DHH: Starship, Starlink and Grok should have been Europe's Sputnik moments — georgemillo · 2026-10-16
- Hinton hits back at Jensen Huang: 'He has good reason to say AI isn't dangerous — he sells Nvidia chips' — BlackHC · 2026-10-16
- Yudkowsky blasts new interpretability finding as primitive; critic pushes back on doomer logic — AlexTensor · 2026-10-16
- Beff Jezos declares 'civilizational inflection point' in the acceleration of technological progress — beffjezos · 2026-10-16
- Data scientist: pivoting to AI engineering is the safest career bet for the next 5 years — mdancho84 · 2026-10-16
- AI will be superhuman at science regardless, and Europe can't opt out — teortaxesTex · 2026-10-16