Blog GitHub is Moving to Usage-Based Billing: Here’s What You Need to Know

As GitHub Copilot moves to Usage‑Based Billing, we break down what’s changing, what it reveals about AI cost, and how organizations can get ahead of the shift.

When GitHub Copilot first joined the enterprise stack, it followed a familiar per‑user pricing model that matched how most teams were buying software at the time. Since then, Copilot has evolved. Teams are working with larger context windows, more powerful models, and increasingly complex, agent‑driven workflows that do far more behind the scenes than early use cases ever required.

That evolution changed how AI consumes compute, and it’s made flat‑rate pricing a less accurate reflection of modern usage patterns. In response, GitHub is making a shift of its own.

On June 1, 2026, GitHub Copilot will move to Usage‑Based Billing (UBB), replacing premium request units with GitHub AI Credits. While this may look like just a billing update at first glance, the change reflects a broader, industry‑wide move toward making AI usage and cost more transparent.

For technology and business leaders responsible for AI‑enabled development, this is less about learning a new invoice format and more about understanding how AI cost behaves at scale — and how to manage it as adoption continues to grow.

Why the change?

The original pricing was built for a simpler time. Early use cases focused on lightweight code suggestions, relatively small prompts, and limited model interaction. That’s no longer the world we’re operating in.

Today’s Copilot usage typically includes:

• Larger context windows
• More powerful models
• Agent-driven workflows that break one request into multiple reasoning steps

Usage-based pricing is GitHub’s way of tying cost directly to the actual work the models are doing, rather than spreading that cost evenly across all users.

Importantly, GitHub isn’t an outlier here. Many leading AI platforms (including OpenAI, Anthropic, and Google) already operate on usage‑based consumption models. GitHub is aligning with an industry standard that’s already well established.

What is Usage-Based Billing (UBB)?

In this UBB model, GitHub AI credits will be consumed based on token usage. Think of "tokens" as the small units of text the AI processes. Each Copilot interaction is measured based on the total tokens processed, which may include input, output, and cached context.

• Input: What you send (your prompts and context)
• Output: What the model sends back
• Cache: Reused context, which is billed at a lower cost than sending entirely new context

These tokens draw from a pool of AI credits, and those credits determine what you’re billed.

From a pricing standpoint:

• Copilot Business remains $19 per user per month and includes $19 worth of AI credits (1,900 credits)
• Copilot Enterprise remains $39 per user per month and includes $39 worth of AI credits (3,900 credits)
• Each AI credit costs $0.01
• Credits are pooled across the full organization, not assigned to individual users
• When the included pool is exhausted, organizations can choose whether to allow additional usage at published rates or cap spend

Organizations still license Copilot per user, but total cost now depends on how usage consumes the shared pool of included credits.

What this means for you

For many organizations with typical usage patterns, the included credit pool will cover most day‑to‑day usage — meaning the impact may not be immediate.

However, the way you manage your budget is going to change:

• From budgeting to forecasting: While Copilot licensing remains fixed, additional spend beyond the included credit pool becomes variable based on actual usage. Predictability doesn’t disappear, but it now depends on visibility, baselines, and ongoing usage monitoring rather than a single renewal number.
• Chargeback models get complicated: Because credits are pooled, teams no longer “own” the usage tied to their licenses. Heavy users and agent-driven workflows can consume a disproportionate share of the pool, even if headcount is evenly distributed. This is where many traditional cost-center assumptions start to break down.
• A small group can drive a big bill: Usage is rarely uniform. A small number of power users (or a handful of complex agent workflows) tend to account for most consumption. Without the right visibility, that imbalance often goes unnoticed until after the fact.
• Developer habits matter: Under flat pricing, efficiency habits didn’t have financial consequences. Under UBB, every prompting decision, model choice, and context size has a measurable cost signal attached to it.

How GitHub is easing the transition

GitHub isn’t asking customers to flip a switch overnight. Instead, it’s rolling out UBB in stages, with clear milestones to help you understand and prepare for the change.

Important dates to circle:

• April 27, 2026: You received a "preview" showing what your April usage would have cost under the new system. We’ll help you use this to set your baseline.

• June 1, 2026: UBB goes live, with usage reflected on the next billing cycle.

The preview window is particularly important. It gives you a chance to establish a baseline before variable invoices begin showing up.

GitHub also shared additional details in its announcement, including temporary promotional credits during the initial rollout, clarification on which Copilot experiences do not consume AI Credits, and updates to how usage is governed once included credits are exhausted. While these specifics are important to review, the broader takeaway remains the same: usage visibility and proactive planning are now central to managing Copilot at scale.

Insight can help you navigate the shift

We don’t treat UBB as a pricing issue. We see it as an operating‑model shift, one that affects how organizations budget for AI, govern usage, and support developers at scale. Our role is to help you adapt to that shift with clarity and confidence.

We start by helping leaders understand what’s actually happening in their environment today. That includes visibility into how GitHub Copilot is being used, where AI credits are being consumed, which models and workflows drive the most activity, and how current usage would translate under UBB. For many organizations, this is the first time they can clearly answer a simple but critical question: “What is AI really costing us?”

From there, we help teams build practical fluency around tokens and usage — because cost control starts with behavior, not policy. Developers learn how everyday decisions like prompt scope, context size, and model selection affect consumption. Engineering leaders gain insight into identifying and coaching heavy usage patterns early. Platform and admin teams learn which configuration levers matter most and when to use them.

Finally, we’ll put the right guardrails in place without slowing teams down. That includes designing layered budgets, setting up proactive alerts, and aligning controls with existing finance and chargeback models. When done thoughtfully, governance becomes a safety net instead of a blocker, leading to more predictable spend while keeping Copilot productive for the people who rely on it most.

The bottom line:

Usage-based billing doesn’t introduce a new cost; it makes existing usage patterns more visible and controllable. Organizations that build visibility, educate their teams, and implement the right controls will often pay roughly what they were paying before, while gaining access to more powerful AI capabilities. Those that wait until after the first variable invoice lands tend to have a much harder conversation with finance.

GitHub has set the timeline. The question for leaders now is whether they want to enter June with a baseline and a plan — or enter July reacting to data they’re seeing for the first time.

Insight is ready to help, phase by phase, whenever you’re ready to start.