GitHub Copilot, the AI code-writing assistant, transitioned to token-based pricing on June 1,
a significant change from the previous fixed premium subscription.
The new model introduces the concept of 'AI credits': developers pay for the number of tokens,
consumed during complex interactions, while the base subscription cost remains unchanged.
GitHub's Chief Product Officer Mario Rodriguez announced this transition on April 27,
citing the growing compute demands of Copilot's agent features as the primary reason for the change.
This move has caused widespread concern among developers.
GitHub community discussion threads about the changes have accumulated over 400 comments and nearly 900 downvotes,
many specialists warn that heavy use of agentic coding features will lead to significant cost increases. One developer calculated that a single Copilot agent session could consume credits worth $30 to $40
— three to four times the monthly limit for Pro plan users.
Under the new system, 1 AI credit equals $0.01, with each request billed based on the number of input, output, and cached
tokens. The cost depends on the selected AI model, request length, and complexity of the generated result. Unused credits do not roll over to the next month.
GitHub is replacing the old extended request units system with AI credits, applicable across all four Copilot pricing plans.
The $10/month Copilot Pro subscription includes $10 in AI credits, while the $39/month Copilot Pro+ plan includes $39 in credits. Business and Business Premium subscriptions are billed per user and provide temporary promotional credits of $30 and $70 respectively, valid until August 2026 for a smooth transition to the new system.
Annual subscription users will continue on the old pricing model until their current contract expires.
It is important to note that two core Copilot features — code autocompletion and next edit suggestions
— remain unlimited and are not subject to the new token-based pricing. Paid mode applies only to chat requests, intelligent workflows, and code review tools.
GitHub representatives stated that this transition is a necessary adjustment reflecting
real-world service usage scenarios. Previously, when load thresholds were exceeded, the platform would switch
to a budget operating mode, which was criticized by developers due to reduced response quality.
The new system allows users to choose their preferred AI models and pay the corresponding token costs.
For the convenience of enterprise customers, GitHub has implemented multi-level budget
control and in May launched a bill preview feature so users can estimate upcoming expenses.