GitHub Copilot is set to move from a flat subscription system to a per-token billing model starting June 2026, aligning its pricing with how large language model APIs are typically charged. Instead of receiving a fixed number of “premium requests” each month, users will now consume “AI credits” based on the number of tokens processed in both input and output. Tokens represent small units of text, meaning that even a single coding query can translate into thousands of billable units depending on complexity. Simple tasks will cost fewer credits, while larger or multi-step coding operations involving extensive code analysis or generation will consume significantly more. The change reflects a broader industry shift, as other AI providers have already adopted similar usage-based pricing structures. While the model offers more precise cost tracking, it may also discourage experimentation and increase expenses for developers and businesses relying heavily on AI-assisted coding tools.
Read more here
https://www.artificialintelligence-news.com/news/per-token-ai-charging-comes-to-github-copilot/
