Anysphere, the company behind the AI-powered coding environment Cursor, recently faced significant user dissatisfaction due to poorly communicated changes to its Pro plan pricing. On July 4, 2025, CEO Michael Truell issued a public apology, acknowledging the company’s missteps in conveying these updates.
The controversy began on June 16, 2025, when Anysphere announced modifications to Cursor’s Pro plan. Previously, subscribers were entitled to 500 rapid responses from advanced AI models such as those from OpenAI, Anthropic, and Google, followed by unlimited slower responses. The new structure allocated $20 worth of usage per month, billed at API rates. This meant users could utilize their preferred AI models until reaching the $20 limit, after which additional credits were required to continue usage.
This change led to confusion and frustration among users. Many reported depleting their allocated requests quickly, sometimes after just a few prompts, especially when using resource-intensive models like Anthropic’s Claude. Others were surprised by unexpected charges, not realizing that exceeding the $20 usage limit without setting a spending cap would result in additional costs. Notably, only Cursor’s auto mode, which dynamically selects AI models based on capacity, offered unlimited usage under the Pro plan.
In response to the backlash, Anysphere committed to refunding users who incurred unforeseen charges and pledged to improve transparency in future communications. The company explained that the pricing adjustments were necessary due to the increased computational demands of newer AI models, which consume more tokens per request for complex tasks. Under the previous plan, Anysphere absorbed these higher costs, but the new structure passes them on to users.
The incident underscores the challenges AI service providers face in balancing cost management with user satisfaction. As AI models become more sophisticated and expensive, companies like Anysphere must navigate the delicate task of implementing pricing strategies that reflect these costs without alienating their user base.
