Starting June 1, 2026, GitHub Copilot will change how it charges users. Instead of a flat monthly fee, users will pay based on the number of tokens their requests consume. This new system aligns Copilot’s billing with how large language model APIs are priced today, mainly used in business plans.

How the New Pricing Model Works

Previously, users had a set number of premium requests each month based on their subscription tier. A complex task that took hours counted as one request, and even simple questions used one. Now, the model shifts to counting tokens, which are units of text processed by the AI. This means each interaction, whether input or output, costs a certain number of tokens.

Tokens are roughly three-quarters of a word. So, a chunk of code or text with 10,000 words could amount to around 12,000 to 13,000 tokens. If a developer feeds Copilot a large codebase for refactoring, that could quickly use up their monthly token allowance. Inputs, prompts, and outputs all count toward the token total, making the billing more granular and reflective of actual usage.

Changes for Users and Pricing Tiers

Even though the pricing per request is changing, the overall subscription tiers remain the same. Instead of a fixed number of queries, users will get ‘AI Credits’ equivalent to their current allowance. For example, a standard Copilot Pro plan at $10 per month will come with 1,000 credits. GitHub states that each credit is worth about one US cent now.

The number of tokens each credit can buy depends on the complexity of the task, the model used, and other factors like cache size and features requested. Simple queries will cost fewer credits, while complex, multi-part requests to advanced models will consume credits faster. This makes it easier for users to understand their usage and costs based on real work done.

Interestingly, some features will stay free. Code completions and suggestions for next edits will not require credits, giving users a free layer of assistance regardless of their usage. This helps soften the impact of the pricing change and encourages continued use of Copilot’s core features.

Industry Trends and Industry Impact

The move to per-token billing is part of a broader industry shift. Companies like OpenAI and Anthropic have already adopted token-based pricing for their enterprise customers. Microsoft, which owns GitHub, is profitable overall and has subsidized Copilot’s costs with revenue from other parts of its business, like software sales.

This change reflects how AI services are evolving toward more precise, usage-based charges. For developers and teams, it means being more aware of how much their prompts and code analysis cost. It also encourages more efficient use of AI tools, as unnecessary or overly complex requests could quickly eat into their credits.

Overall, this update brings GitHub Copilot in line with industry standards and offers a more transparent way to pay based on actual AI usage. While it may introduce some new considerations for users, it also provides opportunities to optimize how they work with AI coding assistants.

Inspired by

• https://www.artificialintelligence-news.com/news/per-token-ai-charging-comes-to-github-copilot/