GitHub Copilot new use based Billing began June 1, replacing the previous flat rate subscription. Developers have reported using up much of their monthly credits in a matter of hours, leading to widespread complaints and some threatening to stop using the product.
Now each request has a dynamic price, depending on the model used, the type of request, the amount of material sent and the complexity of the response.
microsoft announced the change in April, noting that GitHub Copilot has evolved significantly over the past year. The company explained that it now supports more complex agent workflows that require greater computing resources.
What changed with the new use of GitHub Copilot? Billing based on
Previously, GitHub Copilot used a fixed monthly pricing model, charging $10 per month for Copilot Pro and $39 per month for Copilot Pro+. This structure allowed users to make many requests on different AI models for a predictable fee.
Under the new approach, each request consumes credits from a monthly allocation. The cost per application now varies depending on several factors:
- The chosen AI model
- The type and complexity of the request.
- The amount of context sent
- The complexity of the response generated
Microsoft explains that this change aligns pricing with actual usage and costs, and describes it as necessary to deliver a more sustainable and reliable product experience.
Why developers are complaining about Copilot’s new credit system
Response on the GitHub and Reddit user forum has been largely negative, mostly around how quickly credits run out and how unpredictable per-request costs can be.
One developer on the $39 Copilot Pro+ plan shared that he used about 8% of his monthly AI credit allocation in just two hours, estimating that his quota of 7,000 units could be used up in less than two days.
Another user reported spending more than $6 on a single change request, noting that consumption was “impossible to predict.” One Reddit user mentioned that a session using Claude 4.8 to troubleshoot the website used 1,180 credits, which is about 16% of his monthly Pro+ allocation, for results he described as mediocre.
The main concern expressed is that budgeting for AI usage is difficult when a single feature request can consume a significant portion of monthly credits.
Developers opting for alternatives and how Microsoft is responding
Many users have mentioned plans to move their work to other services. Common options include accessing the Anthropic and OpenAI models directly or routing requests through third-party services such as OpenRouter, RooCode, and LM Studio.
One user described a hybrid approach: use the Pro+ allowance until it runs out and then switch to OpenRouter for the rest of the month. OpenRouter operates within the same VS Code interface, offers a broader selection of models, and allows credits to be renewed for up to a year, unlike the GitHub Copilot assignment.
A GitHub spokesperson confirmed the billing update to The Register, stating that usage-based billing is now active. The new pricing model for GitHub Copilot reflects actual usage and includes spending limits, usage dashboards, and model selection options to help users manage their costs.
The company also announced the launch of a higher capacity level called Copilot Max, aimed at users who require more extensive use.
What affected GitHub Copilot users can do to control costs
Developers concerned about the costs of the new model have several options. They can set spending limits in the GitHub Copilot billing dashboard to control monthly charges.
- Monitoring the usage dashboard helps track credit consumption by application and by model. It can also help to choose less expensive models for routine tasks, while higher-cost models are reserved for more complex jobs.
- The new Copilot Max tier may be worth evaluating if usage consistently exceeds the standard allowance. Some may consider accessing the third-party model through services that offer transfer credits or lower prices per request.
GitHub has not announced any changes to the new billing structure in response to recent feedback. The company emphasizes the use of spending limits and usage dashboards as primary tools for managing costs under the metered system.
