After users faced unexpected additional bills, the CEO of the popular coding environment Cursor was forced to apologize, highlighting the growing economic challenge of providing the latest artificial intelligence models.
Introduction: When price surprises users
The CEO of Anysphere, the company behind the popular AI-powered coding environment Cursor, apologized on Friday for a poorly communicated pricing model change to its $20-per-month Pro plan. These changes led some users to complain that they faced unexpected additional costs, sparking a storm of negative feedback on social media and forcing the company to take a step back.
Michael Truell, CEO of Anysphere, said in a blog post: “We recognize that we did not handle this price rollout well and we are sorry for that. Our communication was not clear enough and came as a surprise to many of you.” This incident is not just a communication error, but a window into the complex economic challenges facing tools that rely on sophisticated artificial intelligence models.
Details of the change that sparked outrage
Truell points to an update made on June 16 to Cursor’s Pro plan. Instead of professional users getting 500 fast responses on advanced AI models from OpenAI, Anthropic, and Google, followed by unlimited responses at a slower rate, the company announced that subscribers would now get $20 worth of usage per month, billed at API rates. The new plan allows users to run coding tasks in Cursor using the AI model of their choice until they reach the $20 limit, after which users must purchase additional credits to continue using it.
However, professional users took to social media to voice their complaints in the weeks following the announcement. Many users said they exhausted their requests in Cursor too quickly under the new plan, in some cases after just a few requests when using Anthropic’s new Claude models, which are particularly popular for coding. Other users claimed they were charged unexpected additional costs, without fully understanding that they would be billed extra if they exceeded the $20 usage limit and did not set a spending cap.
The real reason: The exorbitant cost of advanced models
Truell noted in the blog post that Anysphere changed Cursor’s pricing because “newer models can spend more tokens per request on longer-horizon tasks”—meaning some of the latest AI models have become more expensive, spending a lot of time and computational resources to complete complex multi-step tasks. Cursor had been absorbing these costs under its old Pro plan, but is now passing them on to users.
While many AI models have dropped in price, frontier models remain expensive—and in some cases, more expensive than ever. Anthropic’s recently launched Claude Opus 4 model is priced at $15 per million input tokens and $75 per million output tokens. These expenses can creep into AI coding tools, which appear to be getting more expensive across the industry.
Competitive pressure
Cursor has become one of the most successful AI products on the market, reaching over $500 million in annual recurring revenue largely through subscriptions to its Pro plan. However, Cursor now faces intense competition from the AI providers it relies on, while simultaneously trying to figure out how to serve increasingly expensive AI models at affordable prices.
Anthropic’s recently launched Claude Code AI coding tool has been a huge hit with businesses and has likely taken some users away from Cursor in the process. But if Cursor intends to maintain its market-leading position, it cannot stop working with the most advanced model providers—at least, not until its own internally developed models become reasonably competitive.
Conclusion: A difficult balance between power and cost
Cursor’s pricing crisis highlights the growing tension in the AI economy. Users want access to the most powerful models, but these models are expensive to run. Companies building applications on top of these models will have to find transparent and sustainable pricing models that balance delivering value to users with covering their rising costs. Cursor’s apology and refunds to users are a good first step, but the debate on how to price AI power has only just begun.
