The cost of AI is collapsing
When GPT-4 launched in 2023, running it cost a fortune — enough that all but the largest companies couldn't afford to use it at scale. Three years later, equivalent capability costs roughly 95% less. This price collapse is what turned AI from a research demo into something millions of companies actually deploy. The question now is whether the trend continues, plateaus, or reverses as the next generation of frontier models arrives.
Timeline
- March 14, 2023
GPT-4 launches at $30 per million input tokens — expensive enough that running it across a real product was prohibitive for most companies.
- November 6, 2023
GPT-4 Turbo cuts the price to $10 per million tokens, a three-fold drop on the same generation of capability — the first signal that AI prices fall fast even within a single model family.
- May 13, 2024
GPT-4o ships at $5 per million tokens. The frontier becomes multimodal at half the price of the previous frontier.
- December 17, 2024
OpenAI's o1-mini ships at $2.50 per million tokens — reasoning capability now costs less than basic chat did 18 months earlier.
- April 8, 2025
OpenAI's o3-mini brings frontier-tier inference to $1.10 per million tokens. The 36-month price compression continues uninterrupted.
- September 23, 2025
Mid-cycle competition between Anthropic, Google, and OpenAI pushes the floor for frontier-tier inference below $1 per million tokens for the first time.
- May 15, 2026
GPT-5 launches at roughly $1.20 per million tokens — the first upward move at the frontier in 38 months. The new generation is more capable, but reverses the steady decline that defined the prior three years.
Where things stand right now
The 38-month price compression has paused. GPT-5's launch reset the frontier price upward for the first time since 2023 — not by much, but enough to raise the question of whether the era of constantly-falling AI prices is over, or whether competitive pressure pulls prices back down within the next two release cycles.