Cursor vs GitHub Copilot

At a glance

1. Your editor budget now buys a different contract

A renewal priced on the old $20 tier now buys either $8 Plus, which is Grok-only, or $30 Premium, where Claude and GPT tokens bill at API list price on top.

Two questions decide it: do you need non-Grok models, and does your review happen in GitHub PRs?

Cursor told Reuters it runs at about $2.6B annualised revenue; SpaceX bought it in an all-stock deal at a $60B valuation.

Cursor's ownership changed before this renewal decision. Cursor is now a wholly owned SpaceX subsidiary.

• 21 April 2026: SpaceX and Cursor announced a partnership with a $10B payment and a $60B acquisition option.
• 16 June 2026: SpaceX signed the all-stock deal at a $60B valuation.
• 14 August 2026: the deal closed; Cursor joined the SpaceXAI team (SpaceX absorbed xAI in February 2026) to work on Grok, Grok Build, Grok Bot, Grok API and Cursor.

Cursor states its constraint was compute: “We've wanted to push our training efforts much further, but we've been bottlenecked by compute.”

2. Model ownership and repository ownership impose different limits

Cursor's philosophy: the model is the product, so the editor owns the model and the inference bill. Its own price sheet says so — Plus at $8 offers only Grok models, while Premium opens third-party access billed separately at API list price.

GitHub sells the repository, so Copilot's AI lands in PRs and reviews. That forbids giving review away: Free gets agent mode and the CLI, but PR review waits for the $10 Pro tier.

Cursor pushes back on that line — Premium lists code hosting via Origin and AI code review via Bugbot — but the two still compete on different surfaces: Cursor's $8 Plus cannot run Claude or GPT, and Copilot Free cannot review a PR.

3. The meters punish different habits

Cursor's meter punishes third-party model use specifically: no plan includes those tokens, so Claude or GPT work passes through at API list price. Copilot's meter is a single credit pool per user that agent mode, review, cloud and CLI all drain.

When a Copilot seat exhausts its credits, Pro and above buy extra.

4. Model neutrality now needs a renewal check

Teams that chose Cursor partly to stay model-agnostic now buy from a company whose model is Grok. Ownership changed the incentive. It does not prove that every request routes to Grok.

Cursor states that Grok 4.6, the first joint model, scores between 69.9% and 70.8% on CursorBench and 61 on the Artificial Analysis Intelligence Index.

Before renewing, replay representative repository tasks and record the separately billed third-party usage.

• On Premium, compare the separate third-party token charge with the Claude or GPT work your team believes it ran; a near-zero charge means that work did not run on third-party models.
• Do not buy Plus if non-Grok access is a requirement; test Premium's unbundled bill before committing.

5. Copilot Free runs the agent; PR review costs $10

GitHub's counter-argument is reach: Copilot runs in VS Code, Visual Studio, JetBrains, Xcode, Eclipse, Neovim, Zed and on github.com. Copilot Free already supports agent mode and the CLI, with limited usage.

• Copilot in the cloud starts at Pro.
• PR review, file diff review and custom instructions start at Pro.
• Assigning work to Copilot so it creates a pull request also starts at Pro.

A successful CLI trial does not establish that Free can perform your review workflow.

Pro's $15 of monthly credits per user ($10 base + $5 flex) funds agent mode, code review and cloud work alike.

Try the actual handoff from assigned work to reviewed PR before migrating the team.

6. Choose Grok commitment or repository consolidation today

Teams happy on Grok should buy Cursor; teams whose review lives in GitHub PRs should buy Copilot Pro. Cursor's $8 Plus undercuts Copilot Pro by $2, but only if Grok is enough.

• Choose Cursor if Grok meets your workload and third-party usage is optional.
• Choose GitHub Copilot if review and assigned PR work matter more than owning a separate editor workflow.

Moving to Copilot puts custom instructions (.instructions.md) and PR review behind Pro; moving to Cursor puts Claude and GPT behind list-price billing.

If you need Claude inside a known monthly allotment, Copilot Pro+ at $39 includes Opus within $70 of credits; Cursor bills third-party tokens at API list price on top of $30 Premium.

Today, list the workflows that need Claude or GPT and price them on both plans.

Cost arithmetic at volume

These are computed from the published unit prices, not measured. Each row prints its own formula, so you can replace our assumption with your traffic and redo it.

Decision matrix

Rows are the situation you are actually in. Pick the row before you read the pricing page again.

Code and configuration

// Cursor's agent works on the open repository, so the unit of work
// is a task, not a suggestion. The bill follows the model you route to.
// .cursor/rules/backend.mdc — project rules apply to every agent run
---
description: Backend conventions
globs: services/**/*.py
---
Never edit migrations by hand; generate them and review the diff.
// Here it breaks: on Premium, requests routed to Claude or GPT are billed
// at API list price separately from the $30 subscription — watch the invoice,
// not just the seat count.

// Copilot's advantage is the review surface, not the cursor.
// .github/instructions/backend.instructions.md — applied to agent work
---
applyTo: "**/*.py"
---
- Generate migrations, never hand-edit them.
- Every new endpoint needs an integration test.
// Then assign an issue to Copilot: it opens a PR you review like any other.
// Here it breaks: on Free, PR review, diff review and custom instructions
// are unavailable — the agent runs, the review handoff does not.

What you gain, what you give up

Migration path

Common questions

Sources & verification

Prices and limits were read from these pages on 2026-11-10. Vendor marketing claims are quoted as claims. Anything derived is arithmetic on the numbers below.

Keep reading