Reference: Legacy Billing Mechanics, Credit Rates, and Plan Nuances

1. How to Use This Reference

This chapter is a reference, not a lesson sequence. Consult it when you need to look up the retired premium-request billing system, the October 2025 pricing snapshot, exact per-model AI Credit rates, or Free/Student plan nuances — you do not need to read it start-to-finish, and nothing later in this course depends on it.

Chapter 2 already covers the current GitHub AI Credits billing model and the current plan lineup, and that is where you produced two records: a provisional plan recommendation for the reference project team and your own account's live Copilot plan record. This chapter does not repeat that explanation; it only extends it with volatile, legacy, and dated detail that would otherwise clutter early, practice-focused chapters.

Completing this chapter is optional and non-blocking: you can skip it entirely and continue with the rest of the course, and return to it later only if you need to answer a specific 'what was true when' or 'what does GitHub actually publish' question. The four activities below are optional lookup drills, not graded coursework — do any, all, or none of them, and treat each as practice for building a reference note like the ones this chapter models. There is no recommended first-pass drill among the four — each is an equally optional reference lookup, and none is a required or preferred starting point.

2. The Legacy Premium-Request Billing System

Historical model

Before GitHub AI Credits, GitHub Copilot billed most non-completion actions — pull request reviews, Autofix generation, and Coding Agent sessions — against a monthly 'premium request' quota assigned per plan. Each qualifying action consumed one or more premium requests from that pool.

October 2025 quotas (historical)

As of October 2025, the per-plan premium-request quotas were:

These figures are a historical snapshot only — do not treat them as current billing behavior.

Current status: legacy, grandfathered

Per GitHub's own current documentation, premium-request billing is now explicitly labeled a 'legacy' system. It applies only to Copilot Pro and Pro+ subscribers who were on an existing annual plan and remained on legacy premium-request billing after the 2026-06-01 cutover to GitHub AI Credits. Every other current subscriber, on any plan, is billed under AI Credits.

One sentence of recap (Chapter 2 owns the full explanation): AI Credits bill by token usage at published per-model rates rather than by a flat monthly action quota, which is why the premium-request model above is legacy rather than current.

3. October 2025 Plan Lineup and Pricing Snapshot

October 2025 snapshot (historical, superseded)

As of October 2025, GitHub Copilot had five plans, with no tier above Pro+:

Treat this table as a dated historical record, not current pricing.

For today's plan lineup, allowances, and the plan decision you already made for the reference project, see Chapter 2 — that chapter owns the current figures, and this reference does not restate them.

4. Per-Model Rate Table (USD per 1M Tokens) and the Absence of a Typical Cost Figure

Current per-model rates (dated 2026-08-26, subject to change)

GitHub publishes per-model rates in US dollars per 1 million tokens, split into input and output rates. A representative snapshot:

Cached input tokens are typically billed at roughly 10% of the standard input rate. This table changes as GitHub's model lineup changes, so re-check the date whenever you rely on it. Credits convert to dollars at the rate Chapter 2 establishes.

No documented 'typical interaction' cost

GitHub does not publish a fixed AI Credit cost for a 'typical' chat turn, PR review, or cloud-agent session. Only the per-token unit rates above are documented for these interaction types; actual consumption depends entirely on how many tokens a given interaction uses.

Because of that, any 'this costs about N credits' figure you see for a chat turn, PR review, or cloud-agent session — including any illustrative estimate you build yourself in this chapter's activities — is at best an estimate built from the rates above, never a documented GitHub figure. Label any such estimate explicitly, with today's date, as exactly that.

5. Free and Student Plan Nuances

Historical Free-plan limits (superseded)

The Free plan was previously documented with three numeric limits: 2,000 code completions/month, 50 chat messages/month, and 50 premium requests/month.

Current status

Check GitHub's plans overview and its usage-based-billing-for-individuals page for today's Free-plan picture, and expect a mixed update rather than a simple number swap:

• The 2,000-completions/month cap is unchanged.
• Neither page publishes a monthly chat-message cap under any metric name, and neither states a confirmed quantified replacement.
• Free's AI Credit allowance is stated only as 'an allowance,' with no specific published number.

Copilot Student

Current documentation states Copilot Student includes unlimited code completions — distinguishing it from Free's 2,000-completion cap — plus an AI Credit allowance that carries the same unquantified caveat as Free's.

No numeric Free or Student AI-Credit figure appears on those pages. Do not estimate or infer one; state it as unquantified.

Activity 1: Legacy Premium-Request Lookup Drill

Goal: Build a short, status-labeled reference note on the legacy premium-request system.

Steps:

1. Look up how premium-request billing worked historically: a monthly quota per plan, consumed per action for things like PR reviews, Autofix generation, and Coding Agent sessions.
2. Record the October 2025 per-plan quotas — Free 50/month, Pro 300/month, Pro+ 1,500/month, Business 300/month, Enterprise 1,000/month — and label them explicitly 'historical (October 2025)'.
3. Look up premium requests' current status and record it as 'current legacy/grandfathered': scoped only to Copilot Pro/Pro+ subscribers on an existing annual plan who remained on legacy billing after the 2026-06-01 cutover to AI Credits.
4. Write one contrasting sentence using only a brief AI Credits recap — Chapter 2 owns the full explanation — e.g., 'Unlike the flat monthly premium-request quota, AI Credits bill by actual token usage at published per-model rates.'

Expected outputs:

• A note with two clearly labeled sections, 'historical (October 2025)' and 'current legacy/grandfathered,' that never presents premium requests as today's default billing.

Hints:

• If you're not on a grandfathered annual Pro/Pro+ plan, expect not to see premium requests in your own billing at all.

Checkpoint: The historical (October 2025) premium-request quotas and today's current legacy/grandfathered scope should each be labeled with their status, and premium requests should never be presented as today's default billing behavior.

Optional reference-project connection: If you do this drill, check whether your own account's live Copilot plan — not the provisional team recommendation you selected for the reference project in Chapter 2 — is unaffected by, or explicitly scoped under, this legacy system, and record that in your note. That provisional recommendation, as a new selection, would itself fall under AI Credits.

Activity 2: October 2025 Plan Snapshot Card

Goal: Build a dated historical snapshot card for the October 2025 plan lineup.

Steps:

1. Look up the October 2025 lineup and pricing: Free, Pro ($10/month/user), Pro+ ($19/month/user), Business ($19/month/user), Enterprise ($39/month/user) — no individual tier above Pro+ existed at that time.
2. Record it on a card titled 'October 2025 — historical, superseded.'
3. Add one line pointing to Chapter 2 for today's lineup, without repeating those figures here.

Expected outputs:

• A dated 'October 2025 plan snapshot' card kept visibly separate from any current-pricing material.

Hints:

• Do not update or 'correct' these numbers toward current figures — the point is to preserve the historical snapshot intact and clearly dated.

Checkpoint: Your snapshot card should be explicitly dated and labeled historical/superseded, with its figures kept separate from — never merged or averaged with — current pricing.

Optional reference-project connection: If you do this drill, use the card only to contrast against the plan you already selected in Chapter 2; it does not change that decision.

Activity 3: Per-Model Rate Lookup and Bounded Illustrative Estimate

Goal: Read the current per-model rate table and, optionally, build a properly bounded estimate.

Steps:

1. Record today's per-model rates in US dollars per 1M tokens (input/output) for at least three models — e.g., GPT-5 mini ($0.25/$2.00), Claude Sonnet 4.5 ($3.00/$15.00), Gemini 3.6 Flash ($0.75/$3.75, promotional through Dec 2026). Optionally, convert one figure to AI Credits using the rate Chapter 2 establishes.
2. Write today's date next to the table, since GitHub's rate table changes with model lifecycle.
3. Confirm and record that GitHub publishes no fixed 'typical session' AI Credit cost for a chat turn, PR review, or cloud-agent session — only these per-token rates.
4. If you build an illustrative worked example (e.g., a hypothetical 2,000-input/500-output-token exchange), compute it only from the dollar rates you recorded, and label it explicitly: 'illustrative estimate, not a documented GitHub figure, dated [today's date].' Optionally, convert the result to AI Credits using the rate Chapter 2 establishes.

Expected outputs:

• A dated per-model rate table excerpt, plus an optional, clearly disclaimed illustrative estimate that cites only the recorded per-token rates.

Hints:

• Any worked example missing the 'illustrative estimate, not documented' disclaimer violates this chapter's no-fabrication rule — add it. Cached input tokens run roughly 10% of the standard input rate; note this if your estimate involves cached context.

Checkpoint: No 'typical interaction cost' should be asserted anywhere in your output as a documented GitHub fact; any illustrative estimate must carry its dated disclaimer.

Optional reference-project connection: If you do this drill, use the lookup to sanity-check the AI Credit cost implications of model choices on the reference project, using only published unit rates — never a fabricated per-interaction figure.

Activity 4: Free and Student Plan Nuance Comparison

Goal: Compare what changed, what was dropped, and how Student differs — without inventing numbers.

Steps:

1. Look up the historical Free-plan limits: 2,000 code completions/month, 50 chat messages/month, 50 premium requests/month.
2. Check GitHub's current Free-plan documentation (the plans overview and usage-based-billing-for-individuals pages) and record: the 2,000-completions cap is unchanged; neither page publishes a chat-message-cap metric, and no confirmed replacement appears on either of them; Free's AI Credit allowance is stated only as 'an allowance,' with no published number.
3. Check current Copilot Student documentation and record its nuances: unlimited code completions (vs. Free's 2,000 cap), and an unquantified AI Credit allowance carrying the same caveat as Free's.
4. Write a short comparison note stating plainly that no confirmed numeric replacement exists for the dropped Free chat-message cap.

Expected outputs:

• A short Free-vs-Student note that flags the dropped chat-message metric and both plans' unquantified AI Credit allowances as open on current documentation, rather than filled in with invented numbers.

Hints:

• Resist filling in a plausible chat-message or AI-Credit number for Free or Student — write 'unquantified, no confirmed figure' instead.

Checkpoint: No numeric Free or Student AI-Credit allowance, and no replacement chat-message cap, should be asserted anywhere in your output.

Optional reference-project connection: If anyone on your reference-project team is on Free or Student, this drill gives them a non-fabricated picture of their plan's current nuances, without altering the plan decision made in Chapter 2.

Reference Project Continuity — Chapter 7

Starting state: Learners already selected and justified a Copilot plan for the reference project's team scenario in Chapter 2, using only current, non-legacy AI Credits billing concepts. This chapter changes none of that — it adds no reference-project code and does not gate any later chapter.

This chapter's contribution: This chapter is an optional, non-blocking lookup layer over Chapter 2's plan decision. It produces reference documentation rather than reference-project code changes, is not consumed by any later required course work, and exists so volatile/legacy billing detail never displaces the practice-focused chapters that build the reference project.

Ending state: Learners leave with four small, dated reference artifacts they can hold alongside their Chapter 2 decision — a legacy premium-request scope note (Activity 1), an October 2025 pricing snapshot card (Activity 2), a dated per-model rate table with its illustrative-estimate boundary (Activity 3), and a Free/Student nuance comparison (Activity 4). Each artifact is optional documentation, not a required deliverable for the reference project itself, and none of it is consumed by any later required course work — the chapter is a lookup layer over the Chapter 2 decision, not a new build step.

Sources

• The GitHub Copilot Handbook (book)
• https://docs.github.com/en/copilot/concepts/billing/copilot-requests
• https://docs.github.com/en/copilot/reference/copilot-billing/models-and-pricing
• https://docs.github.com/en/copilot/concepts/billing/usage-based-billing-for-individuals
• https://docs.github.com/en/copilot/get-started/plans