What are GitHub Copilot premium requests?
Premium requests were GitHub Copilot's way of metering access to advanced models and features like premium chat and agents. Each plan included a monthly allowance — Pro got 300 — and powerful models counted as 5x or 20x. As of June 1, 2026, GitHub replaced them with AI Credits, where 1 credit = $0.01.
Why — the first-principles explanation
Premium requests existed to solve a cost problem. Basic code completions are cheap to run, so GitHub can include them unlimited on paid plans. But the fanciest models — big-context, advanced-reasoning engines and autonomous agents — cost far more per call. Metering those with 'premium requests' let GitHub give everyone unlimited everyday help while charging heavy users of the expensive features.
Each plan came with a monthly premium-request allowance. In the legacy system, Free included about 50, Pro 300, Pro+ 1,500, Business 300 per user, and Enterprise 1,000 per user. One 'request' was one interaction with a premium model or feature. Crucially, models had multipliers: a standard premium model might cost 1x, while an advanced reasoning model could count as 5x or even 20x per interaction — so a single heavy query could burn through your allowance fast.
The design taught users a habit: match the model to the task. Use lightweight (included) models for routine work, and save premium requests for problems that genuinely need deep reasoning or large context. Once you exhausted your monthly allowance, you either waited for the reset or paid overage per request.
As of June 1, 2026, GitHub simplified this by moving to AI Credits, a dollar-denominated budget where 1 credit = $0.01. Instead of counting requests, plans now include a credit balance (Pro $15, Pro+ $70, and so on) that different models draw from at different rates. Premium requests still appear as 'legacy' billing for some annual plans, but the direction is clear: a single, money-based meter that makes the true cost of each model visible.
An example that makes it click
Think of an amusement park. General admission gets you unlimited rides on the small stuff — the carousel and bumper cars (your basic completions). But the giant new roller coaster needs special tickets: premium requests. A normal thrill ride costs one ticket; the mega-coaster with VR costs five or even twenty tickets. Your plan hands you a book of tickets each month.
In 2026 the park swapped ticket books for a prepaid cash card — AI Credits — where every ride just deducts its real price in cents. Same idea, clearer pricing: you see exactly how much the big coaster costs each time you ride.
How to do it
- Check your plan's allowance: in the legacy system Free ~50, Pro 300, Pro+ 1,500, Business 300/user, Enterprise 1,000/user per month.
- Note that advanced models use multipliers, so one query can cost 5x or 20x a single premium request.
- Use included (non-premium) models for routine work to preserve your allowance or credits.
- As of June 2026, review your AI Credit balance instead, where 1 credit equals $0.01.
- Enable overage or pay-as-you-go only if you knowingly want to exceed your monthly budget.
Key facts
- Premium requests metered access to advanced Copilot models and features beyond basic completions.
- Legacy monthly allowances: Free ~50, Pro 300, Pro+ 1,500, Business 300/user, Enterprise 1,000/user.
- Some models used multipliers, so a single interaction could count as 5x or 20x a premium request.
- Basic code completions did not consume premium requests on paid plans.
- On June 1, 2026, GitHub moved most plans to AI Credits, where 1 credit = $0.01 USD.
The AI pair-programmer built into your editor.
Affiliate link — we may earn a commission at no cost to you.▶ The 60-second explainer (script)
What are GitHub Copilot premium requests? They were Copilot's way of metering the expensive stuff. Here's the logic. Basic code completions are cheap, so paid plans include them unlimited. But the powerful models — big-context, advanced-reasoning engines and autonomous agents — cost a lot more per call. So GitHub gave each plan a monthly allowance of premium requests. Pro included 300, Pro+ included 1,500, Business 300 per user, Enterprise 1,000. One request equals one interaction with a premium model. But watch out for multipliers: an advanced reasoning model could count as 5 times, or even 20 times, a single request — so one heavy question could drain your allowance fast. The lesson was to match the model to the task: use the light, included models for routine work, and save premium requests for problems that truly need deep reasoning. Now, the update: as of June 1st, 2026, GitHub replaced premium requests with AI Credits for most plans. Instead of counting requests, you get a dollar budget — one credit equals one cent — and each model draws from it at its real rate. Same idea, clearer pricing. Premium requests still linger as 'legacy' billing on some annual plans, but credits are the future.
What authoritative sources say
People also ask
How many premium requests did each plan include?
In the legacy system, roughly Free 50, Pro 300, Pro+ 1,500, Business 300/user, and Enterprise 1,000/user per month.
Why do some models cost more than one request?
Advanced models carry multipliers because they're more expensive to run, so a single interaction can count as 5x or 20x a standard premium request.
Do basic completions use premium requests?
No. Standard code completions on paid plans don't consume premium requests; only advanced models and features do.
What replaced premium requests?
As of June 1, 2026, most plans use AI Credits, a dollar-based budget where 1 credit equals $0.01, with each model drawing at its own rate.