Summary created by Smart Answers AI<\/p>\n
In summary:PCWorld reports GitHub Copilot is switching from flat-rate to usage-based AI Credits pricing starting June 1, maintaining $10 Pro and $39 Pro+ monthly costs.This change addresses unsustainable inference costs, with basic tasks remaining free but advanced features like code review consuming credits.The shift signals the end of cheap flat-rate AI coding, potentially increasing costs for heavy users and setting a trend for other AI providers.<\/p>\n
It was fun while it lasted, but it\u2019s starting to look like the end for flat-rate AI plans as we know them, with GitHub being the first to turn out the lights.\u00a0<\/p>\n
Just a week after announcing it was halting signups for its flat-rate Copilot Pro and Pro+ plans, Github has announced that starting in June, those plans will switch over to usage-based pricing<\/a>.<\/p>\n Both GitHub Copilot Pro and Pro+ will still cost $10 a month and $39 a month, respectively, while Business and Enterprise will remain $19 and $39 a month per seat.<\/p>\n But beginning June 1, those plans will replace a fixed allotment of \u201cpremium requests units,\u201d which are based on a user\u2019s AI request count and adjusted based on the strength of the model, with \u201cAI Credits,\u201d which are based on the actual tokens used during AI exchanges.<\/p>\n Under the new plan, for example, Github Copliot Pro users will still pay $10 a month, but instead of getting a set number of PRUs, they\u2019ll get $10 worth of AI credits, while Pro+ users will get $39 worth of monthly AI credits. A similar AI credit allotment will apply for Business and Enterprise users.<\/p>\n While code completion and other basic AI tasks won\u2019t consume AI credits, more advanced and agentic-style activities such as Copilot code review will cost AI credits, GitHub says. Users who spend all their AI credits before the month is up will have the option to buy more.<\/p>\n In a blog post announcing the change, GitHub said that under its current NPU formula, \u201ca quick chat question and a multi-hour autonomous coding session can cost the user the same amount,\u201d and that up to now, \u201cGitHub has absorbed much of the escalating inference cost behind that usage.\u201d<\/p>\n However, \u201cthe current premium request model is no longer sustainable,\u201d the GitHub post said.<\/p>\n What it all boils down to is the end of de facto flat-rate AI pricing for GitHub users, who will now move over to a token-based pricing policy that\u2019s far more punishing\u2013and more realistic, in terms of actual cost\u2013than the NPUs they\u2019ve been consuming.<\/p>\n GitHub\u2019s move to usage-rate pricing is likely a harbinger of things to come for all flat-rate AI users.<\/p>\n The truth is that the flat-rate plans from Anthropic, Google, and OpenAI have long been loss leaders<\/a>, devised to grow their user bases and get new subscribers hooked on their AI-powered wares.\u00a0<\/p>\n Now the big three AI providers are victims of their own successes, particularly after rolling out powerful agentic functionality to their individual consumer plans that burn through tokens at a furious rate.<\/p>\n We\u2019ve already seen Anthropic toy with the idea of dropping Claude Pro and its token-heavy agentic abilities from its $20-a-month Claude Pro plan<\/a>, while Anthropic and competitors OpenAI and Google have been caught silently cutting the usage allotments<\/a> for their flat-rate plans, frustrating subscribers who suddenly found their usage meters running dry.<\/p>\n