Home / Guides / GitHub Copilot Cost
How Much Does GitHub Copilot Cost? (2026 Pricing Breakdown)
GitHub Copilot looks cheap, $10/month for Pro. But the real cost in 2026 hinges on AI Credits, per-token model rates, and the June 2026 switch to usage-based billing. Here's the complete breakdown with worked examples.
Last updated October 2026 · By Soren Starck
GitHub Copilot Pricing at a Glance
Copilot has five paid tiers plus a Free plan in 2026. Every paid tier gives you unlimited code completions, then meters chat, agent mode and other model calls in AI Credits (1 credit = $0.01) at each model's per-token rate against a monthly allowance.
| Plan | Price | Code completions | AI credits / mo |
|---|---|---|---|
| Free | $0 | 2,000 / month | An allowance, auto model only (50 chat cap) |
| Pro | $10/mo ($100/yr) | Unlimited | 1,500 ($15) |
| Pro+ | $39/mo | Unlimited | 7,000 ($70) |
| Max | $100/mo | Unlimited | 20,000 ($200) |
| Business | $19/seat/mo | Unlimited | 1,900 / user, pooled |
| Enterprise | $39/seat/mo | Unlimited | 3,900 / user, pooled |
The headline price tells you almost nothing about your real cost. Two developers both on Pro can have wildly different bills under the June 2026 usage-based billing. The variable that matters is AI Credits: which models you use and how many tokens they burn.
How AI Credits Are Charged
An AI credit is a flat $0.01, but what one ask costs is not. GitHub bills each interaction by the model and the number of tokens, at the rates in its models and pricing table:
| Model / Task | Per 1M tokens (input / output) | Notes |
|---|---|---|
| Code completions | Not billed | Unlimited on paid plans, no credits used |
| GPT-5 mini | $0.25 / $2.00 | Lightweight |
| Gemini 3.8 Flash | $0.75 / $3.75 | Promotional rate through December 31, 2026 |
| Claude Sonnet 5.5 / GPT-6.1 Sol | $2.00 / $10.00 | Versatile and powerful models |
| Claude Opus 5.5 | $4.00 / $20.00 | Heavy reasoning model |
| Agent / multi-step task | Sum of every call | One ask, many model calls |
This is the trap. On Pro you get 1,500 credits ($15), which sounds like plenty until you realize a single long agent run on Opus can eat several hundred of them. Two heavy agent sessions a week and your monthly allowance is gone.
Annual Copilot Pro and Pro+ subscribers who stayed on the legacy request-based billing still spend premium requests with model multipliers until their plan ends; GitHub lists those in its legacy multiplier table.
The June 2026 Change: AI Credits & Usage-Based Billing
Before June 2026, hitting your premium allowance meant Copilot silently fell back to the base model, your autocomplete just quietly got worse. As of June 1, 2026, GitHub replaced premium requests with AI Credits and usage-based billing:
- Included allowance first, your plan's monthly AI Credits (1,500 on Pro, 7,000 on Pro+, 20,000 on Max, 1,900 or 3,900 per seat on Business or Enterprise) come with the subscription.
- Then overage at $0.01 per credit, billed at each model's per-token rate: a chat turn that sends 20K tokens and gets 1K back costs about 5 credits on Claude Sonnet 5.5 and about 10 on Claude Opus 5.5; a long agent run can cost hundreds.
- You set a budget, overage is opt-in. The default budget is $0, so you must explicitly set a budget for additional usage or you simply stop at your included allowance.
- Org-level controls, Business and Enterprise admins can cap or disable overage per seat.
The upside: no more silent degradation if you opt in. The downside: your bill is now variable. A $10 Pro plan can quietly become a $40+ plan in a heavy month if you let overage run unmonitored. This is exactly why tracking matters now more than it did a year ago.
Worked Examples: What You'll Actually Pay
Autocomplete-first developer (Pro, $10)
You mostly use inline completions (no credits, unlimited) and reach for chat a few times a day on Claude Sonnet 5.5. Roughly 600 credits ($6 of tokens) a month, under your 1,500 allowance.
Code completions: unlimited, $0
~600 credits of Sonnet chat: within 1,500 allowance, $0 overage
Effective monthly cost: $10
Agent-heavy developer (Pro, $10 + overage)
You run Copilot agent tasks daily, often on Opus. Say 3,900 credits ($39 of tokens) a month: 1,500 included, 2,400 over.
Subscription: $10
2,400 overage credits × $0.01 = ~$24 overage
Effective monthly cost: ~$34 (close to Pro+ at $39, which includes 7,000 credits)
This is the crossover insight: once your overage approaches $29/month, you're better off on Pro+ ($39, $70 of credits included). Without tracking, most developers never notice they crossed that line.
Team of 10 (Business, $19/seat)
Ten developers at $19/seat is $190/month base. Each seat adds 1,900 credits to a shared pool, so the team gets 19,000 a month. If three of them are agent-heavy and drain the pool, the org bill climbs unpredictably unless admins set budgets. Use the plan calculator to model whether Business or Enterprise (3,900 credits per seat) is cheaper for your team's burn rate.
SessionWatcher
Your Copilot bill is now variable.
See the burn before the invoice.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
Overage Cost: The Numbers
Under June 2026 usage-based billing, extra usage costs the same per-token rates once your included allowance is gone. Some typical asks:
| Request type | Tokens (input / output) | Approx overage cost |
|---|---|---|
| GPT-5 mini chat | 20K / 1K | ~$0.007 (under 1 credit) |
| Claude Sonnet 5.5 chat | 20K / 1K | ~$0.05 (5 credits) |
| Claude Opus 5.5 chat | 20K / 1K | ~$0.10 (10 credits) |
| Agent task (Opus 5.5, uncached) | 1M / 100K | ~$6.00 (600 credits) |
A single careless week of Opus agent runs can add $15–$30 in overage to a $10 plan. The default $0 budget protects you, but the moment you enable overage to avoid degraded autocomplete, you need eyes on the meter.
How to Track Your Copilot Spend
There are two practical ways to know what you're spending:
1. The GitHub billing dashboard: shows AI Credits usage, a per-model breakdown and overage with a delay. It's accurate but reactive, you find out you blew your budget after the fact, not while you're working.
2. SessionWatcher: SessionWatcher for Copilot tracks your premium request count, burn rate, model breakdown, projected month-end overage, and billing-cycle countdown live in your Mac menu bar or Windows taskbar, with notifications at 70% and 90% by default. You see the dollar cost climb as you work, so you can pace yourself or switch to the base model before overage bites.
SessionWatcher Pricing (Not the Same as Copilot's)
To be clear: the prices above are GitHub's charge for Copilot. SessionWatcher is a separate, one-time purchase that monitors that spend, it does not replace your Copilot subscription.
Copilot monitoring is included with SessionWatcher Pro
$79 one-time or $24/year, covers all supported tools
One licence key, up to 5 devices: any mix of Macs and Windows PCs
14-day refund, no subscription required
SessionWatcher is $79 once. If it stops $50 a month of unnecessary Opus overage, it has paid for itself within two months; it also shows you when you've outgrown Pro and should move to Pro+. See SessionWatcher for Copilot.
If your company pays for Copilot Business seats and wants each developer to see their own burn, SessionWatcher Pro for Teams is $40 per seat per year (or $4 per seat per month, 3-seat minimum), one Pro licence per teammate on one invoice. It covers Macs only for now, and it is not an org dashboard: each developer sees their own usage on their own machine, so org-level budgets still live in GitHub.
Is GitHub Copilot Worth It?
For autocomplete-first developers, Copilot Pro at $10/month is one of the best-value AI coding tools available, unlimited inline completions alone justify it. The risk is purely on the AI Credits side. If you lean heavily on agent runs and reasoning models, the variable overage can quietly multiply your bill, and at that point Pro+ ($39) or a dedicated Claude/Codex plan may be more economical.
The right answer depends entirely on your AI Credits burn rate, which you can't guess. Track it with SessionWatcher or model it with the plan calculator before you commit.
SessionWatcherStop guessing your Copilot bill.
$79 once, no subscription.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
Frequently Asked Questions
How much does GitHub Copilot cost per month?
Copilot Pro is $10/month ($100/year), Pro+ is $39/month, Max is $100/month, Business is $19/seat/month, and Enterprise is $39/seat/month. There's also a Free plan with hard monthly caps. Every paid tier includes unlimited code completions plus a monthly AI Credits allowance: 1,500 on Pro, 7,000 on Pro+, 20,000 on Max, 1,900 per user on Business and 3,900 per user on Enterprise.
What changed with Copilot billing in June 2026?
GitHub replaced premium requests with AI Credits and usage-based billing on June 1, 2026. Once you exhaust your included allowance, extra usage is billed in AI Credits (1 credit = $0.01) at each model's per-token rate against a budget you set, instead of silently dropping to the base model. Overage is opt-in with a default $0 budget. Model multipliers now apply only to legacy annual Pro and Pro+ plans.
How much do Copilot premium requests cost?
Premium requests have been replaced by AI Credits (1 credit = $0.01), except on legacy annual Pro and Pro+ plans that still use multipliers. Usage is billed per token: Claude Sonnet 5.5 is $2 / $10 per million input / output tokens, Claude Opus 5.5 $4 / $20. A typical Sonnet chat turn costs about 5 credits, an Opus turn about 10, and an agent task can run several dollars. Code completions do not use credits on paid plans.
How can I track my Copilot spend?
SessionWatcher tracks your premium request burn rate, model breakdown, and projected overage live in your Mac menu bar or Windows taskbar. Copilot monitoring is included with SessionWatcher Pro ($79 one-time, or $24/year, covering all supported tools), with a 14-day refund. The GitHub dashboard also shows usage, but with a delay.