GitHub Copilot Users Face AI Billing Shock

GitHub Copilot users are experiencing considerable sticker shock as the service's new usage-based AI credit pricing model takes effect. Many users have reported that their previous normal usage now rapidly consumes their allotted monthly AI credits, with some exhausting their quotas in under a day.

The shift from request-based billing to a credit system has surprised many. Under the new model, paid subscriptions grant users a set number of AI credits monthly, with each credit valued at $0.01. Subscription tiers offer varying credit amounts: $10/month Pro plan includes 1,500 credits, $39 Pro+ offers 7,000 credits, and $100/month Copilot Max provides 20,000 credits.

The cost per prompt is determined by input and output tokens, alongside the chosen large language model. This means pricing is highly variable; for instance, one million output tokens could cost $1.25 on one model but $30 on another. Some complex prompts or extended coding sessions have been reported to consume thousands of credits, leading some users to threaten subscription cancellations.

While many users are contemplating alternative AI coding solutions, some have found ways to adapt. By limiting AI usage to focused, deliberate changes, users can manage their credit consumption effectively. The industry may see a broader adoption of similar usage-based pricing, potentially favoring AI models that are more token-efficient.