GitHub Copilot’s New Usage-Based Pricing Model Sparks Backlash From Developers

Microsoft’s GitHub Copilot is facing growing criticism after announcing a major change to its pricing model, moving away from flat-rate subscriptions and introducing a usage-based system tied to token consumption. While larger organisations may be able to absorb the additional costs, many individual developers and smaller teams worry that the new structure could significantly increase their monthly expenses.

The changes, which took effect on June 1, mean users will now be billed based on the number of tokens consumed while using Copilot’s AI-powered features rather than paying a predictable subscription fee tied primarily to request limits.

The announcement quickly triggered backlash across developer communities on platforms such as Reddit and X, where users shared concerns about what they believed could be dramatic cost increases.

One Reddit user described the new pricing structure as unreasonable, claiming that their monthly bill could rise from roughly $29 to nearly $750 under the updated model. The developer argued that at those rates, the service would no longer provide enough value to justify the expense and said they planned to cancel their subscription.

Another user posted screenshots suggesting their projected costs had jumped from around $50 per month to nearly $3,000, expressing shock at the scale of the increase.

Not everyone agrees with the criticism, however. Some developers have argued that such extreme costs are likely the result of unusually heavy usage patterns rather than normal software development work.

Several users suggested that developers who understand how to use AI coding assistants efficiently should not be generating enough token usage to produce bills of that magnitude. According to these critics, the largest increases are likely affecting so-called “vibe coders,” users who rely heavily on AI-generated code and repeatedly run large numbers of prompts, revisions, and iterations.

One developer noted the significant difference between users reporting minimal overage costs and those sharing screenshots showing thousands of dollars in projected charges. The commenter argued that excessive reliance on AI-driven coding workflows is likely responsible for the most dramatic memes circulating online.

Others focused on a different question entirely: how sustainable GitHub Copilot’s previous pricing model had been in the first place.

Several users speculated that Microsoft may have been heavily subsidising usage under the older flat-rate structure. Given the computational costs of running advanced AI models, some developers questioned how the company could continue to support high-volume users without adjusting pricing.

The economics behind AI coding assistants have long been difficult to evaluate from the outside, and the actual costs Microsoft incurred supporting intensive Copilot usage remain largely unknown.

At the same time, some developers believe Microsoft bears responsibility for the backlash. Critics argue that the company actively encouraged users to integrate Copilot deeply into their workflows and to take advantage of increasingly powerful AI features. As a result, they say many customers built habits around a pricing model that no longer exists.

One user argued that developers were using the product exactly as Microsoft designed and promoted it. According to that view, the problem is not how customers used Copilot but rather the company’s decision to replace a predictable subscription model with a token-based system that can generate significantly higher costs for heavy users.

The debate highlights a broader challenge facing the AI industry. As AI-powered tools become more capable, the infrastructure required to operate them grows increasingly expensive. Companies must balance providing powerful services with maintaining sustainable business models, while users continue to expect predictable and affordable pricing.

For now, GitHub Copilot’s pricing overhaul has become a flashpoint within the developer community, with supporters viewing the change as a necessary adjustment and critics seeing it as a costly shift that could make advanced AI coding assistance far less accessible for many users.