GitHub Copilot moved every plan to usage-based billing on June 1, 2026 — instead of a flat per-seat fee covering unlimited use, Pro now includes $15 in monthly AI credits ($10 base plus $5 "flex"), and Pro+ includes roughly $70. Unlimited code completions and the cloud agent are still included outright; it's the more advanced, premium-model requests that now draw down your credit balance. The catch that caught teams off guard: unused flex credits don't roll over, resetting to zero at the start of each month, and a temporary September 2026 promotional-credit boost for Business and Enterprise plans expired right on schedule — a real drop in available usage for anyone who hadn't checked the fine print.
What actually changed, and what didn't
Copilot's core autocomplete and chat experience inside the IDE is unchanged — it's still the most widely deployed AI coding assistant specifically because it's built directly into the editors and repos teams already use, rather than asking them to adopt a new tool. What changed is exclusively the billing model: Pro is $10/month with unlimited code completions plus $15 in monthly credits for premium-model requests (the base $10 is fixed, the extra $5 "flex" allotment is described as variable and may change), and Pro+ is $39/month with premium models included and a larger roughly $70 monthly credit allowance (about 7,000 credits). Business and Enterprise get pooled per-user allowances of 1,900 and 3,900 credits respectively.
The September 2026 credit cliff
Business and Enterprise plans got a temporary promotional boost — 3,000 and 7,000 credits respectively — running from June 1 through September 1, 2026, as a cushion during the transition to usage-based billing. That promotional window closed exactly on schedule, dropping allowances back to the standard 1,900 and 3,900. Teams that scaled their usage up during the promotional period without checking that it was temporary saw a real, sudden capacity drop in September — worth flagging to your engineering org now if no one has specifically confirmed your team's current allowance against actual monthly usage.
Is it still worth it?
For teams primarily using standard code completion and chat, yes — the $10/month Pro tier's unlimited completions cover the large majority of daily usage, and the $15 credit allowance is enough for occasional premium-model requests without careful rationing. The calculation changes for teams that lean heavily on premium models (GPT-5-class or Claude-class requests) for complex refactors or planning — that usage draws down credits quickly, and Pro+'s larger allowance, or a completely different tool priced for heavy agentic use, may be the more honest fit. Compare directly against Claude Code (from $20/month, usage counted against unified Claude limits) and Cursor if premium-model usage, not basic completion, is your team's actual bottleneck.
Quick comparison
| Plan | Price | What's unlimited | Monthly credits |
|---|---|---|---|
| Pro | $10/mo | Code completions, cloud agent | $15 ($10 base + $5 flex) |
| Pro+ | $39/mo | Completions + premium models | ~$70 (~7,000 credits) |
| Business | Per seat | Completions | 1,900/user (pooled) |
| Enterprise | Per seat | Completions | 3,900/user (pooled) |
FAQ
Do unused Copilot credits roll over to the next month? No — credits (including the flex portion) reset to the full monthly allowance at 00:00 UTC on the first of each month, and anything unused is forfeited rather than carried forward.
Why did my team's Copilot usage suddenly feel more limited in September 2026? Business and Enterprise plans had a temporary promotional credit boost from June 1 to September 1, 2026, cushioning the transition to usage-based billing — that boost expired on schedule, dropping allowances back to standard levels, which can feel like a sudden cut if no one flagged it as temporary.
Does the $5 "flex" credit portion ever change? GitHub describes the flex allotment as variable and subject to change over time, unlike the fixed $10 base — check current plan details directly rather than assuming it's permanently fixed at $5.
Is Copilot still cheaper than Cursor or Claude Code for basic use? For pure autocomplete and chat within a supported IDE, Copilot's $10/month Pro tier remains one of the lowest entry prices in the category — the cost gap widens specifically once premium-model usage becomes a regular part of your workflow.
Related guides
- Claude Code vs Cursor vs Codex vs GitHub Copilot in 2026
- Cursor vs GitHub Copilot
- How Much Do AI Coding Agents Cost in 2026?
- Explore Coding & Dev tools
*News and pricing details reviewed as they change. Last updated September 2026.*
Bogdex · Founder & editor, woska
Bogdex builds and curates woska, testing AI tools against real workflows to judge which ones actually save time rather than which have the longest feature list.