per AI Credit. The new atomic billing unit.
your plan dollar equals your monthly credit wallet.
code completions stay free on all paid plans.
worldwide auto-migration from PRUs to UBB.
Same price. Different unit of measurement.
Your Copilot subscription cost doesn’t move. What moves is how usage is counted — from opaque “premium requests” to per-token AI Credits, visible per model and per interaction.
Before June 1, 2026
- Fixed premium request units (PRUs) per plan
- Silent fallback to a cheaper model when limits are hit
- Annual plans with auto-renew available
- Flat model multipliers — no per-token transparency
After June 1, 2026
- Monthly GitHub AI Credits wallet equal to plan dollars
- No fallback — credits govern every call to every model
- Monthly billing only — annual plans retired
- Per-token pricing visible per model, per interaction
Base price is unchanged. Your plan dollars become your wallet.
Across Free, Pro, Pro+, Business, and Enterprise — base subscription rates hold. Each paid dollar becomes one dollar of monthly AI Credits.
$39 base + flex allotment. Best for power users running agents and frontier models.
Best ValuePooled credits across the organization with per-user spend controls.
Team RolloutPooled credits with four-level budget hierarchies and full governance.
Governed ScaleCompletions remain free. Agents and chat now cost tokens.
Code completions and Next Edit Suggestions remain unlimited on every paid plan and are never billed in AI Credits. Heavier surfaces — chat, agents, code review — consume credits proportional to the model and the work.
Six moves that protect your credits without slowing the team.
Token efficiency is now a first-class engineering practice. These are the habits we install when we run our hands-on workshop.
Write specific prompts
Vague queries trigger longer exchanges and waste tokens. Specificity compresses cost.
Right-size the model
Use lighter models for routine work. Reserve frontier models for genuinely complex problems.
Curate context, don’t dump it
Pasting large files into chat is billed as input tokens. Reference, summarize, or scope down instead.
Watch the meter
GitHub provides real-time credit tracking in Billing Overview. Make it a daily glance, not a monthly surprise.
Set per-user budgets
Org admins can cap spend per seat. A $0 user budget disables Copilot for that seat — useful for graceful offboarding.
Keep tooling current
Update your IDE and Copilot extensions so usage alerts and per-model pricing stay accurate.
Three engagements. Every stage of UBB readiness.
From a single leadership briefing to a two-week enterprise rollout — pick the depth of partnership your team needs to land on the new billing model with confidence.
Executive Webinar
- GitHub Copilot UBB overview
- AI Credits & consumption model
- Governance & adoption planning
- Cost optimization strategies
- Enterprise readiness guidance
- Engineering best practices
1-Day Hands-on Workshop
- UBB fundamentals & AI credit model
- Prompt engineering best practices
- Context engineering & guardrails
- Token optimization techniques
- Real-world coding scenarios
- Prompt reuse & compression strategies
2-Week Engineering Enablement
- Baseline usage assessment
- Copilot customization
- Organization instructions & standards
- Productivity engineering
- Model sizing & optimization
- Architecture-aware prompting
- AI consumption governance
- Adoption acceleration framework
Official docs and Canarys field notes.
We’ve published deep dives on AI Credits, cost impact, token efficiency, and agent customization. Start here.
Land the new billing model before it lands on you.
We help engineering teams optimize AI usage, install budget controls, and build efficient Copilot workflows ahead of June 1.
