Billing switch: from PRU to AI Credits
GitHub Copilot moved all plans to AI Credits usage billing starting June 2026, with 1 AI Credit = $0.01. Monthly Pro/Pro+ subscribers migrated automatically on June 1, while annual subscribers stay on the old PRU multiplier until expiry (source).
The essence of this overhaul is replacing a vague allowance with transparent credits. Pro/Pro+ used to bill through PRU (Premium Request Unit) multipliers that were hard to calculate; now everything is unified into Credits at one cent each, with Chat, Agent, PR review and premium models drawn from Credits while code completions and Next Edit Suggestions stay free.
Old vs new billing
| Dimension | Old (PRU) | New (AI Credits) |
|---|---|---|
| Unit | PRU multiplier | 1 Credit = $0.01 |
| Transparency | Hard to calculate | Simple per-cent billing |
| Completions | Counted | No Credits |
| Overage | Hard stop/downgrade | Continues at token rates |
Tier-by-tier pricing comparison
The pricing comparison is simple: higher tiers include more Credits and unlock more models. The list prices across individual and team tiers are (source):
| Tier | Price | Included Credits | Positioning |
|---|---|---|---|
| Free | Free | 2,000 completions + limited Chat | Trial |
| Pro | $10 | 1,500 (1,000 base + 500 flex) | Most popular |
| Pro+ | $39 | 7,000 (3,900 + 3,100) | Power |
| Max | $100 | 20,000 (10,000 + 10,000) | Max usage |
| Business | $19/seat | ~1,900/seat | Teams |
| Enterprise | $39/seat | ~3,900/seat | Enterprise |
The pricing comparison conclusion: individuals should start with Pro, then move to Pro+ or Max if they routinely downgrade models or stop agent tasks because credits run low. Team tiers pool credits per seat into a shared pool where admins can set a budget cap.
How AI Credits are actually spent
Not everything costs Credits: code completions and Next Edit Suggestions are always free—what really costs money is Chat, Agent tasks, PR review and premium model calls. Once you exceed the monthly allowance there's no hard stop; usage keeps billing at each model's token rate (source).
Copilot's catalog packs 28 models, including OpenAI GPT-5.6, Claude Opus 5, Gemini 3.6 Flash, Kimi K3 and Grok 4.5. Different models carry different token rates—Opus 5 clearly costs more than Haiku 4.5, which is where the biggest Credits gaps appear.
Is upgrading worth it after the switch?
After moving from PRU to AI Credits, whether upgrading is worth it depends on your real monthly usage and whether you need premium models. If you mostly use completions and occasional Chat, Pro's 1,500 Credits is enough; if you run Claude Code-style agents daily and frequently call Opus or GPT-5.6, Pro+'s 7,000 Credits keeps premium models on by default.
Annual users should note: annual billing continues on the old PRU multiplier until expiry, then moves to Free or converts to a monthly plan early with prorated credit for the remaining value. If your usage would overrun under the new billing, converting early and upgrading usually saves money.
Things to watch
-
1 Credit = $0.01, but the flex allowance changes. Base credits are fixed 1:1 to your subscription price, while the flex portion adjusts over time—don't treat it as permanent.
-
Annual migration has a window. Monthly already switched June 1; annual switches at expiry, so decide before then whether to go Free or upgrade to a monthly tier.
-
Team tiers are shared pools, not per-person. Business/Enterprise credits pool at the org or enterprise level with admin budget caps, so individual allowances don't apply.
-
Overages don't stop but keep billing. There's no hard cutoff, but token-rate overage billing means a runaway agent can push your bill past expectations.