OpenAI's recent strategy to drastically reduce prices for its AI models appears to be paying off, as falling costs have led to an unexpected but spectacular increase in revenue. According to an analysis by TD Cowen, based on data from the OpenRouter platform, cheaper access to technology did not reduce turnover; instead, it skyrocketed demand.
The Numbers Behind the Growth
OpenAI implemented an 80% price cut for the GPT-5.6 Luna model and a 20% cut for the mid-tier Terra model. The results were striking:
- For Luna, the effective usage price dropped about 10 times, while consumption increased 14 times, resulting in a 34% revenue increase.
- For Terra, the price fell 3 times, but usage quintupled, leading to a 45% boost in revenue.
This growth is calculated in comparison to the seven-day period prior to the implementation of the new price lists, indicating that customers reacted immediately to the lower cost per token.
The Jevons Paradox in AI
This development is a modern application of the "Jevons Paradox." In the 19th century, economist William Stanley Jevons observed that when the use of a resource (such as coal) becomes more efficient and cheaper, its total consumption increases rather than decreases. In the case of AI, low costs allow businesses to expand its use into new areas, such as mass document analysis, customer service automation, and the operation of complex AI agents.
Competition and Future Prospects
Aggressive pricing also seems to be affecting market share. Data from Ramp shows that OpenAI’s GPT-5.6 Sol model managed to capture a larger share of enterprise spending in July than Anthropic's Fable 5, likely due to its more attractive pricing.
Despite the optimism, TD Cowen analysts remain cautious, as the data covers only two weeks of usage. It remains to be seen whether this explosive revenue growth will be sustained in the long term or if it represents a temporary market reaction.
