GitHub Copilot

What the price actually is

Individual Pro plan billed monthly; an annual rate is advertised lower. Free, Business and Enterprise tiers are priced separately and are not covered by this entry.

The usage this is priced against

A developer using it most working days: about 4,000 inline completions and 150 chat turns a month. Completions are priced against a small fast model, chat against a mid-tier one.

Someone using it half as much sees double the multiple. The assumption is the argument — if you disagree with it, the number below is not about you. Change it in the calculator.

The cost math, in full

Rates come from a dated rate card of representative published API prices, not from the vendor. Nobody outside these companies knows what they actually pay; volume discounts and in-house serving both push real costs below these figures, which makes every multiple here a floor rather than a ceiling.

Why this verdict, not the number

The estimated inference cost lands within a few percent of the sticker price, which means essentially none of the $10 is markup on tokens at this volume. Whatever the plan is buying — the editor extensions, the context-gathering that makes completions relevant, the request routing — is not being funded by a margin on the model. Lighter users subsidise heavier ones, which is how a flat fee works, but the flat fee itself is not the interesting number here.

What the price buys besides tokens

• Integrations
• Scale & infra
• Distribution

The hard part is not calling a model, it is deciding what to send it: which files, which symbols, how much of the buffer, fast enough that the suggestion arrives before you have typed past it. That retrieval and latency work is the product. The IDE extensions across VS Code, JetBrains, Neovim and Visual Studio are a long maintenance tail that no individual replicates.

What you lose if you leave

Sub-second inline completion inside your editor, the context selection that makes those completions relevant, and extensions maintained against several IDEs that each break on their own schedule.

The honest cheaper path

Continue.dev or Tabby wired to an API key, or a local model through Ollama for completion-only use. Both are real options and both leave you owning the context-gathering and the latency tuning.

Recompute it for yourself

Where every number came from

• pricingGitHub Copilot plans
• docsGitHub Copilot documentation
• rateRate card entry: frontier-small
• rateRate card entry: frontier-mid

Price recorded 11 Aug 2026 · entry last reviewed 11 Aug 2026. Think something here is wrong? File a correction — we publish them, including the ones that embarrass us.