What GitHub’s Token-Based Billing Means
GitHub Copilot’s new token-based billing model is a usage-based pricing system where developers pay through metered AI Credits that are consumed according to the number of tokens processed for each request rather than a flat subscription. This change replaces the earlier blend of fixed monthly subscriptions and premium request units that loosely governed high-intensity features such as advanced chat and long-running agentic sessions. GitHub now assigns each plan a pool of GitHub AI Credits, and every interaction with supported models draws from that pool based on published token rates for inputs, outputs, and cached data. Code completions and next edit suggestions remain credit-free, but anything involving heavy model use now has a clear cost. The result is tighter alignment between developer subscription costs and actual consumption, but also more volatility in monthly bills, especially for those who depend on long, autonomous coding sessions.
From Premium Requests to an AI Credits System
Under the old model, GitHub Copilot bundled a fixed subscription with premium request units that covered expensive operations without tightly tying them to cost. As agentic workflows grew longer, GitHub absorbed soaring inference bills, a situation the company now calls unsustainable. The new AI Credits system replaces those units with a monthly credit allowance denominated in tokens. Plan prices stay the same, but each tier receives a mix of base credits and, for individuals, a flexible top-up that GitHub can adjust as model economics change. According to The New Stack, “Pro users (USD 10, approx. RM46 per month) receive USD 15 (approx. RM69) in total monthly credits, and Pro+ users (USD 39, approx. RM180 per month) receive USD 70 (approx. RM322).” Once credits are exhausted, additional usage becomes metered, making the cost of heavy features transparent instead of hidden inside a flat monthly fee.
New Copilot Max Plan and Individual Cost Impacts
For heavy users, GitHub has introduced Copilot Max, a top-tier individual plan priced at USD 100 (approx. RM460) per month that includes USD 200 (approx. RM920) in AI Credits. Max provides USD 100 (approx. RM460) in base credits plus an equal “flex allotment,” aimed at developers who run sustained, high-volume agentic sessions and want to avoid constant credit ceilings. Existing Student, Pro, and Pro+ subscribers can move to Max, though new sign-ups for individual plans are temporarily paused. In practice, many users report that the switch to token-based billing has raised their effective GitHub Copilot pricing. One commenter cited by ArtificialIntelligence.News said their “12% of total AI credits burned like anything for very minor task,” while another showed using nearly half of a 7,000-credit allowance in a single day. Light autocomplete users may be fine, but intensive chat and agent workloads now carry a clear price tag.
Budget Controls and Flex Allocation for Organizations
Business and enterprise customers keep per-seat prices at USD 19 (approx. RM87) and USD 39 (approx. RM180) per user per month, respectively, with matching AI Credit allotments. Unlike individual plans, these tiers do not get a flex allotment; credits are matched 1:1 with license cost, though GitHub is offering temporary promotional boosts through August. The key structural change is pooled credits at the organization level, letting power users consume more while lighter users offset them. A new budget control stack adds user-level caps, overrides, cost center budgets, and an enterprise-wide metered budget. The enterprise budget does not cap total monthly spend; it only limits metered overage once pooled credits run out. If user-level budgets collectively exceed the shared pool, excess usage spills into metered territory, and a “lowest remaining headroom wins” rule can cut off some users earlier than their individual caps if the enterprise budget is set too low.
