Artificial intelligence usage by businesses is reaching new highs, but the cost of consuming AI models is moving in the opposite direction. According to the latest Ramp AI Index update, average AI token prices fell 21.2% over six weeks, while token consumption climbed to a record level. The findings highlight how rapidly the economics of the AI industry are changing as companies adopt AI tools and providers compete on pricing.
The data also reveals a growing gap between AI usage and spending. Token volume increased even as the amount businesses paid for AI services declined from its recent peak. For AI companies such as OpenAI and Anthropic, this creates a major challenge: they must convert rapidly growing usage into sustainable revenue while dealing with falling prices and increasing competition.
AI Token Prices Drop 21.2% in Six Weeks
The latest figures were shared by Ara Kharazian, lead economist at Ramp, on November 19, 2026.
According to the updated Ramp AI Index, average token prices declined by 21.2% over six weeks. During the same period, token consumption increased by 1.5% to reach a new record high.
AI token spending also declined by 6.3% from its peak recorded two weeks earlier.
The figures suggest that businesses are consuming more AI-generated output and using AI models more frequently, but the cost of each unit of usage is becoming cheaper.
Ramp’s index draws on financial and AI usage data from more than 85,000 U.S. companies. Its token measurements use usage and billing information imported from AI providers.
Key Figures From the Latest Ramp AI Index
| Metric | Latest Finding |
|---|---|
| AI token price decline | 21.2% in six weeks |
| Token volume | Increased 1.5% to a record high |
| Token spending | Down 6.3% from its recent peak |
| Token volume growth | 27.2% between October 18 and November 15 |
| Token spending growth | 12% over the same comparison |
| Combined average price | $0.45 per million tokens for OpenAI and Anthropic |
The longer comparison is particularly revealing. Between the weeks ending October 18 and November 15, token volume increased 27.2%, while spending increased only 12%.
This means usage is expanding considerably faster than the amount of money businesses are spending on AI APIs.
Why Are AI Token Prices Falling?
Several developments are contributing to the decline in AI token prices.
One important factor is competition among AI model providers. OpenAI, Anthropic, Google and other developers are competing to attract businesses and developers that consume large volumes of AI tokens.
Lower prices can make their products more attractive to customers, particularly companies operating AI agents, coding assistants and automated customer-support systems.
Another factor is the availability of cheaper and more efficient models. Improvements in model architecture, inference infrastructure and computing efficiency can reduce the cost of generating responses.
Open-weight models and new competitors are also increasing pricing pressure. Businesses can increasingly choose between different models depending on their requirements for speed, accuracy and cost.
The result is a market in which customers can potentially process more AI workloads without increasing their budgets proportionately.
OpenAI and Anthropic Face Growing Price Competition
The Ramp data provides a useful comparison between two of the largest AI providers.
Between October 18 and November 15, the combined average price for OpenAI and Anthropic tokens fell to approximately $0.45 per million tokens, declining 11.2% from the previous month.
| AI Provider | Average Price per Million Tokens |
|---|---|
| OpenAI | $0.38 |
| Anthropic | $0.50 |
| Combined average | $0.45 |
These figures represent averages across the models and usage mix observed in Ramp’s data. They are not official, like-for-like API price lists for individual models.
The market-share figures also show how concentrated business AI spending remains.
For the week ending November 15, Anthropic accounted for 58.7% of tracked token spending, while OpenAI represented 37.2%. Google accounted for 1.4%, and xAI represented 2.4%.
Anthropic’s strong position in token spending demonstrates the importance of enterprise demand for its AI products. However, the broader market is becoming increasingly competitive as providers seek to attract customers through better performance and lower costs.
AI Adoption Continues to Expand Across Businesses
Falling prices are occurring alongside growing business adoption.
According to Ramp’s October 2026 figures, 56.8% of tracked U.S. companies paid for AI products or services during the month.
This represented an increase of one percentage point from September and 13.3 percentage points compared with the same period a year earlier.
Anthropic had an adoption rate of 44.8%, while OpenAI reached 40.6% among companies tracked by Ramp.
The figures suggest that AI is becoming a regular business expense rather than remaining limited to technology companies experimenting with new tools.
Companies are using AI for software development, research, customer support, document processing and other operational activities.
However, adoption is not equally distributed across departments.
Engineering Teams Still Drive Most AI Spending
Ramp’s November 11 analysis found that engineering departments accounted for 72% of classified business AI spending in October.
Customer success and support operations represented 13.3%, compared with just 0.8% in June.
This indicates that AI adoption is beginning to expand beyond software engineering, although engineering remains the dominant source of enterprise spending.
The expansion into customer support and other business functions could create new opportunities for AI providers. These workloads may also favor faster and cheaper models rather than the most expensive frontier systems.
Why Rising AI Usage Does Not Automatically Mean Higher Revenue
The biggest challenge for AI companies is that higher consumption does not necessarily translate into proportionately higher revenue.
Consider a company that previously spent $100 on AI services. If token prices fall significantly, the same budget could support substantially more usage.
The business might generate more reports, process more documents or run additional AI agents without increasing its spending.
This is beneficial for customers, but it can put pressure on the revenue generated per token by AI providers.
The industry must therefore find ways to increase total demand faster than prices decline.
Possible opportunities include:
- More businesses adopting AI across additional departments.
- AI agents performing longer and more complex tasks.
- Higher demand for advanced reasoning and specialized models.
- Greater integration of AI into enterprise software.
- New consumer and business applications that generate additional usage.
The challenge is particularly important because AI companies and infrastructure providers are investing enormous amounts in data centers, GPUs and computing capacity.
Those investments require sustained demand and revenue growth to justify their costs.
The Bigger Picture
The latest Ramp AI Index highlights an important transition in the AI economy. The industry is moving from a period in which access to powerful AI models was expensive toward one in which businesses can consume substantially more AI at lower unit costs.
This could accelerate adoption across industries, making AI-powered automation accessible to smaller businesses that previously faced high operating costs.
However, falling prices also mean that model providers must compete harder to maintain revenue growth. The long-term winners may not necessarily be the companies charging the highest prices, but those capable of delivering reliable AI services at scale while maintaining sustainable economics.
Looking Ahead
The next major indicator to watch will be whether AI token consumption continues growing quickly enough to offset declining prices. If businesses expand their use of AI agents, coding assistants and automated workflows, total demand could continue increasing even as the cost of individual tokens falls.
For OpenAI, Anthropic and other AI developers, the priority will be converting that growing consumption into recurring enterprise revenue. The latest figures show that AI adoption is still expanding, but the economics of the industry are becoming more competitive as customers receive more computing power for every dollar spent.
SEO Title: AI Token Prices Fall 21% as Usage Hits Record High, Ramp Finds
Meta Description: Ramp AI Index finds AI token prices fell 21.2% in six weeks while usage hit a record, highlighting growing demand and pricing pressure on AI companies.
Suggested URL Slug: ai-token-prices-fall-21-percent-ramp-ai-index
Focus Keyword: AI token prices
Secondary Keywords: Ramp AI Index, AI token usage, OpenAI pricing, Anthropic AI, AI industry spending, AI model costs, enterprise AI adoption
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.



