AI coding assistants · Comparison
GitHub Copilot vs Cursor vs Claude Code: Copilot leads at 8.3; 5 developers cost $95 to $150 a month
We read the pricing, docs, trust and legal pages of three AI coding assistants on the same day and scored them on the same six indicators. What 5 developers cost, how usage is counted, what the terms say about your code, and who should pick which.
On this page9 sections
On six indicators read from each vendor's public pages, GitHub Copilot scores 8.3 out of 10, Cursor 6.4 and Claude Code 5.6. For our reference scenario, 5 developers on the cheapest organization plan with central billing, a no-training commitment, and chat plus an agent mode with usage included, Copilot Business comes to $95 a month, Claude Team Standard seats to $100 on annual billing and Cursor Teams to $150. Copilot stands out for unlimited completions, usage priced in dollars and a wide model choice; Cursor for an agent-first editor with privacy mode enforced for the team; Claude Code for a terminal and IDE agent with Claude chat on the same seat and uncapped IP indemnity.
Copilot is in the Excellent band, Cursor in the Good band and Claude Code in the Fair band. The scores measure what the public pages commit to, not how good the suggestions are or how fast the tools run, and we did not sign up for any of them.
At a glance
| GitHub Copilot | Cursor | Claude Code | |
|---|---|---|---|
| What you pay for | Completions, chat and agents across many editors | An AI-first editor built around agents | A coding agent for the terminal, IDE and web, with Claude chat |
| Free option | Free plan: 2,000 completions a month, limited chat and agent | Free plan with limited Agent and Chat | None with Claude Code; trials only on sales-assisted Enterprise |
| Reference scenario (5 developers) | $95 on Business, billed monthly, computed | $150 on Teams, monthly price, computed | $100 on Team Standard, annual, computed |
| Included usage per seat | 1,900 AI credits ($19), pooled | Premium weekly pool (4x Plus), not pooled | 1.25x Pro per session, shared with chat |
| Models | Anthropic, OpenAI, Google, xAI and more | Cursor's own plus Anthropic, OpenAI, Google | Anthropic only |
| Inline completions | Unlimited on paid plans | Tab; limits not stated | None |
| IP indemnity | Business, with the public-code filter on | In the MSA; plans not stated | Team, no liability cap |
| SCIM and audit log | Business | Enterprise | Enterprise (20-seat minimum) |
| Score (public pages) | 8.3 Excellent | 6.4 Good | 5.6 Fair |
What the public pages show
GitHub Copilot sells Business at $19 per user a month, billed monthly, since annual plans are no longer sold. Each user brings 1,900 AI credits ($19 of chat and agent usage) a month into a pool shared across the organization, completions are unlimited, and usage beyond the pool is billed at per-token rates. Business lists no training on your data, IP indemnity, content exclusion, SAML SSO, SCIM through managed users and an audit log.
Cursor is now sold within The X Subscription. A Teams Premium seat costs $30 a month and carries the Premium weekly pool for Cursor's own models; Anthropic, OpenAI and Google models are paid from prepaid credit at API list price plus $0.25 per million tokens. Teams enforces Privacy Mode for every member, with zero data retention agreements with model providers, and includes SAML SSO.
Claude Code comes with every paid Claude plan. A Team Standard seat costs $20 a month billed annually, or $25 monthly, and covers Claude Code and Claude chat from one usage pool. Team content is not used for training by default, the Commercial Terms give IP indemnity without a liability cap, and Team includes SAML SSO and spend limits.
Strengths and things to know
GitHub Copilot. Strengths: unlimited completions; usage priced in dollars and pooled; models from six providers in many editors; IP indemnity, SSO, SCIM and an audit log on Business. Things to know: extra usage is on by default for organizations and organization budgets stop it only with a setting that is off by default; monthly billing only; prompts from surfaces other than the IDE are kept for 28 days; several pages about the Free plan and trial disagree.
Cursor. Strengths: privacy mode enforced for the team; zero data retention with model providers; SAML SSO on Teams; SOC 2 Type II, ISO 27001 and ISO 42001. Things to know: usage is stated only as multiples of Plus; SCIM, the audit log and pooled usage are on Enterprise, priced by sales; it is sold within a bundle with Grok and X; six pairs of pages disagree or leave a price out, including the annual Teams price.
Claude Code. Strengths: an agent in the terminal, IDEs, desktop, web and CI; chat and agent on one seat; IP indemnity without a cap; admin-managed settings for files, models and MCP servers. Things to know: usage is stated only as multiples of Pro; Anthropic models only and no inline completion; no free plan includes it; SCIM, audit logs and zero data retention need Enterprise.
Pricing reality
Reference scenario: 5 developers on the cheapest organization plan a 5-developer team can buy today with central billing and seat management, a stated commitment that the organization's code and prompts are not used for training, and chat plus an agent mode with usage included in the seat price, billed annually where an annual price is offered, at US list prices before tax, using only the usage the plan includes.
| Service | Plan | Monthly cost | Where the number comes from |
|---|---|---|---|
| GitHub Copilot | Business | $95 | Computed: 5 × $19, billed monthly (no annual price is sold); includes 9,500 pooled credits ($95) |
| Claude Code | Team, Standard seats | $100 | Computed: 5 × $20 on annual billing; $125 billed monthly |
| Cursor | Teams, Premium seats | $150 | Computed: 5 × $30; no annual price is stated for Teams seats |
Month to month, the same seats cost $95 on Copilot, $125 on Claude and $150 on Cursor (computed). The seat price is not the whole cost for heavy agent use. Copilot states its allowance in dollars, so a team can see when the $95 pool runs out; Cursor and Claude state theirs only as multiples of another plan. Beyond the allowance, all three bill at per-token rates, for example $4 in and $20 out per million tokens for Claude Opus 5.5, with Cursor adding $0.25 per million tokens on Teams. Heavier Cursor and Claude users can move to bigger seats: Cursor Super at $100 and Ultra at $200, Claude Premium at $100 a seat on annual billing.
How we scored them
Every product in this category is scored on the same six indicators, read from the vendor's own pricing, docs, trust and legal pages, with fixed formulas. The overall score is the plain mean. Our methodology page explains how these scores work.
| Indicator | GitHub Copilot | Cursor | Claude Code |
|---|---|---|---|
| Pricing clarity | 5 | 2 | 5 |
| Free plan and trial | 8.5 | 8.5 | 1 |
| Price for 5 developers | 7.9 | 6.3 | 7.7 |
| Usage terms and model choice | 10 | 6 | 4 |
| Code privacy and IP protection | 9 | 8.3 | 9 |
| Compliance and admin | 9.3 | 7.2 | 6.9 |
| Overall | 8.3 | 6.4 | 5.6 |
Pricing clarity loses a point for each price a page leaves out or each pair of pages that disagree: five for Copilot, six for Cursor and three for Claude. Cursor and Claude also lose points because they state included usage only as multiples of another plan. Claude Code scores 1 on the free plan item because the Free plan does not include it and trials are listed only for sales-assisted Enterprise. Copilot's lead comes from its usage terms and admin controls: dollar-priced usage, unlimited completions, several model providers, and SCIM and an audit log on the $19 plan. Claude's lower usage score reflects usage stated only as multiples, Anthropic-only models and no inline completions.
Also in this category
JetBrains AI scores 5.3: unlimited completion and usage at API rates with no surcharge inside JetBrains IDEs, with per-user credit limits on the free Air Governance Basic tier, while no organization plan includes usage, so the price item cannot be computed, and there is no IP indemnity. Gemini Code Assist scores 4.1: daily request quotas with no overage, IP indemnity and no prompt storage at $19 a user on annual terms, but Google stopped selling new subscriptions on Oct 9, 2026 and points new customers to Antigravity.
Who should choose which
- Choose GitHub Copilot if your team uses several editors and wants unlimited completions, usage priced in dollars and IP indemnity, SSO, SCIM and an audit log on a $19 seat. Look elsewhere if you want an annual price; if you choose it, set budgets before rollout, since extra usage is on by default for organizations.
- Choose Cursor if you want an AI-first editor with agents, privacy mode enforced for everyone and zero data retention with model providers. Look elsewhere if you need SCIM or an audit log without an Enterprise contract, or want usage stated in requests or dollars.
- Choose Claude Code if you want a coding agent in the terminal or your IDE, with Claude chat on the same seat and IP indemnity in the standard team terms. Look elsewhere if you want inline completions or models from other providers.
Bottom line
From public pages alone, GitHub Copilot leads at 8.3: $95 a month covers 5 developers with unlimited completions, $95 of pooled usage, several model providers and the main enterprise controls. Cursor, 6.4, costs $150 a month and gives the team an agent-first editor with privacy mode enforced, but its usage is stated only as multiples and its pages disagree in several places. Claude Code, 5.6, costs $100 a month on annual billing and suits teams that want a terminal agent and chat on one seat with Anthropic models. Compare the usage terms and admin controls you need, not only the seat price.
How we checked
We read the vendors' pricing pages and plan tables, docs and help centers, trust and security pages and legal terms on Oct 16, 2026. All three showed US dollar prices; for Cursor's X Subscription terms, which refused our request, we read the Internet Archive copy of Oct 13, 2026. We did not sign up, start a trial or buy a seat. Facts the pages did not state, such as the size of Cursor's and Claude's usage allowances or Cursor's annual Teams price, are left out rather than guessed. Every figure is on the vendors' pages as of that date or, where marked, computed from them.
