OpenRouter: a 25 000 percent increase in token consumption does not mean comparable growth in AI usage
Weekly token consumption on the OpenRouter platform has increased by more than 25 000 percent since January 2025, from 0.5 to 126.2 trillion. According to The Decoder, this is not driven by corresponding growth in AI usage, but primarily by thinking tokens from reasoning models and unoptimized agentic systems.
The OpenRouter platform, which developers use to connect AI models to their own products, has recorded an increase in weekly token consumption of more than 25 000 percent since January 2025 – from 0.5 trillion to 126.2 trillion tokens per week. According to The Decoder, however, this chart does not reflect comparable growth in actual AI usage or business value.
According to the article, the sharp increase is primarily driven by reasoning models, which produce large amounts of so-called "thinking" tokens before generating a response, and by unoptimized agentic AI systems, which consume tokens very quickly. A small increase in actual usage can therefore lead to a much larger increase in token consumption.
According to the article, the GPT 5.6 Luna model from OpenAI recently led in the volume of tokens consumed on OpenRouter, but the authors say this does not necessarily mean that more people are using it – the model may simply generate more tokens per prompt. In terms of revenue, however, the Astra model from OpenAI leads. According to the article, the Chinese models Kimi, GLM and DeepSeek are growing rapidly, with monthly spending ten times higher in 2026, although they are starting from a significantly lower base.
Why it matters
The volume of tokens consumed is often treated as an indicator of the popularity of AI models and the growth of the industry as a whole, but according to the article, it can be misleading – the same number of queries can produce many times more tokens with reasoning models or agentic systems without a corresponding increase in actual usage or revenue. For companies paying for tokens, this means a risk of rising costs without a corresponding benefit if their agentic systems are not optimized, and for market observers, it is a warning against judging growth in AI adoption solely by token volume.
Two audiences, two different impacts
What this means
For individuals
High token consumption by a model does not necessarily mean that more people are using it – the model may simply generate more tokens per query, making this an unreliable indicator of actual popularity.
For a business
Companies paying for tokens should not confuse rising token consumption with growth in usage or revenue – according to the article, thinking tokens from reasoning models and unoptimized agentic systems are the main contributors to inflated figures, which may mean rising costs without a corresponding benefit.
StrategyCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.