$29 to $750 per month: GitHub Copilot switches to token-based pricing, developer community erupts

📄Full Article· Automatically extracted by trafilaturaGemini 翻譯1482 words

GitHub Copilot will switch from a fixed monthly fee to token-based billing starting June 1. Some developers estimate their bills will skyrocket from $29 per month to nearly $750. The official community discussion thread has already accumulated nearly 900 downvotes. (Previous coverage: xAI rushes to push Grok Build to v0.2.11! Poaching two core members from Cursor in an attempt to catch up with Claude Code) (Background: AI stock guru Serenity: Dell surges 28%, upstream supply chain has yet to catch up) The new system, effective June 1, shifts Copilot’s billing logic from a "fixed monthly subscription" to "token-based billing." Every feast has a moment when the bill arrives. Microsoft-owned GitHub Copilot is preparing to hand this bill to developers. The community's reaction is more direct than any analysis. In the official GitHub community announcement thread, there are over 400 comments and nearly 900 downvotes—a figure that speaks for itself. A Reddit user wrote in a post that they currently pay about $29 per month, but estimate their bill will approach $750 under the new system, declaring they would cancel their subscription after leaving the comment, "What a joke." The key to why costs are so difficult to control is the agentic coding workflow. Copilot is no longer just completing a few lines of code; it is planning tasks, breaking them down into sub-steps, and continuously executing multiple iterations in the background, with every step consuming tokens. In a single complex refactoring session, the AI must repeatedly read existing code, generate modification plans, execute tests, and adjust based on the results. The total token consumption for this entire loop is far higher than the sum of individual completions. It is not uncommon for a single session to consume $30 to $40, but the Pro plan only provides a $10 monthly allowance, meaning users often hit the limit before a single workday is over. This is why some in the community are asking, "How much money was Copilot actually losing before?" However, the community is not entirely one-sided. Some senior developers hold a different stance: the responsibility for the massive bills lies with the users, not the pricing mechanism. They point out that those who truly understand the tool shouldn't be burning so many tokens, and those seeing massive spikes are mostly "vibe coders" engaged in "vibe coding." The logic among this group is: "If you treat Copilot as a tool rather than a magic wand, the pricing is actually reasonable for small teams." This perspective is not without merit, but it avoids a more fundamental question: who packaged AI as a product experience that encourages users to "use it as much as possible"? GitHub and Microsoft’s marketing over the past few years has never been about teaching users to moderate token consumption; instead, it has been about demonstrating how Copilot can complete entire functional modules for you and refactor thousands of lines of code in one go. The fixed-rate subscription model itself was a form of behavioral inducement, psychologically conditioning developers to treat compute costs as zero and fostering a habit of heavy reliance on agentic workflows. If "vibe coders" are the root of the problem, the answer to who designed this problem is self-evident. Microsoft’s shift in pricing has its own financial logic: the compute cost of agentic coding workflows is orders of magnitude higher than traditional code completion, and the fixed-rate model can no longer sustain gross margins as usage behaviors shift rapidly. From a broader perspective, this is a transition happening across the AI tool industry. After the subsidy period ends, usage-based billing is the inevitable destination for almost all platforms providing compute-intensive services. Copilot is not the first, and it won't be the last. But the problem is not the change itself; it is the timing and the pace. Part of the developers' dissatisfaction stems from "unpredictability." The greatest value of a fixed monthly fee was never just the low cost, but the ability to plan budgets and use the tool with peace of mind. Once switched to token-based billing, users must mentally calculate costs every time they trigger an agentic workflow, immediately increasing the psychological friction of using the tool. For individual developers and small teams, the impact of this change is particularly acute; for enterprises that have deeply integrated Copilot into their CI/CD pipelines, the cost of re-evaluating their toolchain is equally significant.

Data Status✓ Full text extractedRead Original (動區 BlockTempo)

🔍Historical Similar Events· Keyword + Asset Matching6 items

2026-04-24

GitHub Copilot halts self-service subscriptions: AI usage spirals out of control, the economics of affordable plans have completely collapsed

Similarity 140%關鍵字 copilot/github同分類 zh

2026-05-26

Two months after Claude Code burned through Uber's annual budget, the COO bluntly stated: There is no proportional relationship between Token consumption and useful output.

Similarity 120%關鍵字 token同分類 zh

2026-05-26

Microsoft Copilot Cowork hit by major vulnerability: AI Agent automatically leaks confidential enterprise files when faced with prompt injection attacks

Similarity 120%關鍵字 copilot同分類 zh

2026-05-24