Almanac
Microsoft/copilotMicrosoft 365

Consultant KB for the whole Microsoft Copilot estate. Spans Microsoft 365 Copilot, Copilot Studio, the Dynamics 365 Copilots, Power Platform Copilot, Fabric & Power BI Copilot, Copilot in Azure, Windows & consumer Copilot, extensibility & agents, Security Copilot, and GitHub Copilot, plus admin/security and cross-product decision guides. Populated by the daily author agent from Copilot release notes / release plans plus the author's own consultant notes.

feature-github-copilot-plans.mdv2 · history
CurrentApplies to GitHub CopilotUpdated 18 hours agoSource docs.github.com

What it does

Sets out the Copilot tiers, what each includes and how usage is metered. Since June 2026 all plans run on usage-based billing with a monthly allowance of GitHub AI Credits, replacing the old premium-request counters.

Key facts

  • Individual tiers (as of mid-2026): Free (no charge), Student (free, verified), Pro $10/month, Pro+ $39/month, Max $100/month.
  • Organisation tiers: Business $19 per granted seat/month, Enterprise $39 per granted seat/month. Enterprise needs GitHub Enterprise Cloud.
  • Free is deliberately capped: 2000 completions/month, limited chat, and models via auto-selection only — no free model picking.
  • Every paid plan bundles a monthly pool of GitHub AI Credits; higher tiers get larger allowances and priority access to newer models. Business includes roughly 1,900 credits/user, Enterprise roughly 3,900.
  • Billing moved to usage-based on 1 June 2026. Spend beyond the included credit pool is metered at $0.01 per AI credit; premium requests remain only as a legacy reference.
  • On Business and Enterprise the credits are pooled across the enterprise — each licence contributes its allowance to one shared pot rather than a per-user ration. One heavy agent user draws from everyone's.
  • Code completions and next edit suggestions aren't billed in credits at all and stay unlimited on every paid plan. Credit burn comes from chat, agents and the coding agent.
  • Pro is free for verified teachers and maintainers of popular open-source projects.
  • Business and Enterprise add centralised policy control, seat management and the contractual data terms (no training on your data, IP indemnity with duplication detection on).
  • Max is the individual heavy-use tier — Pro+ features plus the largest personal credit allowance and priority model access.

When to use / skip

For a solo developer, Pro at $10 is the obvious entry point; only step up to Pro+ or Max if they are running sustained agent workloads and burning through credits. For any team of more than a handful, go straight to Business — the admin plane and data terms are the reason, not the per-seat price. Enterprise is worth it when you need the github.com chat over your own repos, enterprise policy inheritance and priority model access; otherwise Business covers most needs.

Configuration decisions

  • Business vs Enterprise: decide whether you need enterprise-wide policy inheritance and Copilot on github.com, or whether org-level Business is enough.
  • Set an AI credit / spending limit so agent-heavy users don't run up unexpected metered charges.
  • Decide who gets seats and how — direct assignment, teams, or automatic for all members.
  • Choose whether to allow overage beyond the included credit pool, or hard-cap it.

Gotchas

  • The June 2026 shift to AI Credits changed how cost is reasoned about. Old guidance framed in "premium requests" is out of date — think in credits and metered overage now.
  • Pro+ is $39/month, the same headline number as Enterprise per seat, which trips people up. They are different products (individual vs org-managed).
  • Enterprise requires GitHub Enterprise Cloud. If the client is on Enterprise Server or plain Team, Enterprise Copilot isn't available.
  • Free's caps make it a trial, not a rollout. Don't scope a team pilot on Free and expect representative behaviour.

Consultant notes

  • For business cases, model cost as seats plus expected credit overage from agent use, not just the sticker per-seat price. Agent and coding-agent runs consume credits faster than completions.
  • The pooled credit model is the bit clients don't expect. Budget is set per seat, consumption isn't, so a small group running the coding agent hard can empty the pool for a department that only ever uses completions — and completions are free, so those users never show up in the usage conversation. Agree a monitoring cadence and a spending cap in the same meeting you agree the seat count.
  • The data and IP terms — not features — are usually the reason to insist on Business/Enterprise over individual plans in a corporate setting.
  • Verify current pricing on the day. GitHub has re-cut the plan line-up more than once (Max and the credits model are recent); anything hard-coded here can drift.

Confirm the live per-seat and credit numbers on the GitHub plans page before quoting a client — these get re-priced.

Was this accurate?