GitHub Copilot Embraces Usage-Based Pricing
The era of unlimited flat-rate AI coding assistance may be coming to an end. GitHub has announced that starting June 1, its Copilot Pro and Pro+ plans will adopt a usage-based pricing model, replacing the current system of premium request units with AI credits based on actual token usage.
Under the new structure, Copilot Pro subscribers will still pay $10 per month, but instead of a fixed number of premium request units, they will receive $10 worth of AI credits. Similarly, Pro+ users will get $39 in monthly credits for their $39 subscription. Business and Enterprise plans will follow the same token-based approach, though seat prices remain unchanged at $19 and $39 per month respectively.
What Changes for Users
Basic tasks like code completion will continue to be free and not consume any credits. However, more advanced and agentic features—such as Copilot code review, autonomous coding sessions, and complex multi-turn interactions—will now draw from the user's credit balance. Those who exhaust their monthly credits can purchase additional ones.
GitHub explained in a blog post that the current premium request model was no longer sustainable. “A quick chat question and a multi-hour autonomous coding session can cost the user the same amount,” the post noted, adding that the company had been absorbing escalating inference costs. The shift to token-based pricing is designed to better align costs with actual usage.
Why the Change?
The move reflects a broader industry trend among AI providers. Anthropic, Google, and OpenAI have long offered flat-rate plans as loss leaders to build user bases. But with the rollout of powerful agentic functionality that burns through tokens rapidly, these plans have become financially unsustainable. Anthropic has already hinted at dropping Claude Pro’s agentic abilities from its $20-per-month plan, while all three have quietly reduced usage allotments, frustrating subscribers.
Anthropic’s Head of Growth, Amol Avasare, acknowledged that “AI agents that run for hours weren’t a thing” when cheap flat-rate plans were introduced. The current plans were simply not built for such intensive usage patterns.
What This Means for the Future
Usage-based pricing is more transparent and fair, but it will almost certainly cost heavy users more than flat-rate plans did. An intermediate step might be keeping flat-rate plans for basic chat while charging per token for premium features like code assistants and desktop coworking.
GitHub’s decision could be a harbinger for the entire AI industry. As more providers adopt token-based models, the days of all-you-can-eat AI coding for a fixed price may soon be over. For GitHub users, the change arrives in just a few months—and the bill is likely to rise.
Source: PCWorld News