GitHub has announced that its AI coding assistant, Copilot, will soon charge users based on their actual usage. This shift from fixed request quotas to an AI credit system is designed to better reflect the true cost of computing resources.

The new model aims to address discrepancies in pricing where a quick question and lengthy coding sessions could incur the same costs. Under this updated system, subscribers will receive monthly AI credits linked to their subscription fee, with additional usage priced according to token consumption.

While simple code completion tasks remain free of credit expenditure, more complex features such as code reviews will come at an extra cost in GitHub Actions minutes. The pricing varies significantly depending on the model's sophistication, with OpenAI’s GPT models ranging from $4.50 per million output tokens to $30.

This move reflects a broader trend in AI services adapting to rising costs and increasing demand for computational resources. For developers reliant on Copilot, it may mean a more accurate reflection of the effort required but also more financial oversight over AI usage.