Direct answer: public list prices for enterprise AI coding assistants in 2026 fall into three bands. GitHub Copilot individual plans start at USD 10/month, while GitHub Enterprise is USD 21/user/month and includes Copilot. Cursor Teams is USD 30+/user/month, with third-party models billed separately at API list price. Claude Pro is USD 20/month (USD 17 annual) and includes Claude Code; Max and Ultra are USD 100/200 per month. For 20 developers, visible annual seat cost is roughly USD 2,400–9,600, before model usage, admin/SSO, security, training, and rework. Seat price is a screening input, not a ranking.
This guide is for engineering leaders, platform/DevOps, procurement, finance, and security teams evaluating AI coding assistants. Data cut-off: 31 October 2026. Prices come from official GitHub, Cursor, and Anthropic pricing pages. OpenAI Codex/ChatGPT Business did not provide a verifiable public price during this check, so it is not included in the comparison. All prices are in USD, excluding tax; enterprise agreements, reseller discounts, minimum commitments, and invoices control. This is public-source research and an editorial model, not a quote, and it does not promise productivity gains, savings, or a go-live date.
Seat price is not TCO. Usage, credits, and third-party models create a second bill. If your existing code platform already provides identity, audit, and exit controls, do not rebuild them for one tool.
What are the official AI coding-assistant seat prices in 2026?
GitHub, Cursor, and Claude all offer individual plans and team/enterprise tiers. Individual plans are cheaper but lack central administration; enterprise tiers cost more because they add SSO, audit, data boundaries, and consolidated billing.
Separate individual subscriptions from enterprise procurement. Individual plans are monthly, tied to a personal account, and suitable for trials. Enterprise plans are per seat and add organisation management, SSO/SCIM, audit, data retention or residency options; prices often require a sales quote. The table lists only public list prices verified during this check and does not guess custom enterprise prices.
| Vendor / plan | Official price (USD) | Included / fit | Procurement note |
|---|---|---|---|
| GitHub Copilot Pro | 10 / month / user | Individual; chat, agent, CLI, completions | No org SSO/audit; extra AI Credits sold separately |
| GitHub Copilot Pro+ | 39 / month / user | Higher AI Credits and advanced features | Still an individual plan, not enterprise governance |
| GitHub Copilot Max | 100 / month / user | Highest individual usage tier | For heavy individuals, not a team default |
| GitHub Enterprise | 21 / month / user | Includes Copilot; repository platform, SSO, audit, data residency | Not a pure Copilot price; standalone Copilot Business quote not obtained |
| Cursor Plus | 8 / month | Grok models and basic AI usage | Grok only; third-party models extra |
| Cursor Premium | 30 / month | 4× Plus, cloud agents, code review; third-party models at API price | Third-party usage not included |
| Cursor Super / Ultra | 100 / 200 / month | 15× / 50× Plus usage | Individual tiers; team admin is Teams |
| Cursor Teams | 30+ / month / user | Team management, SSO, usage analytics | “+” means higher tiers may exist; third-party models extra; pooled usage is Enterprise |
| Claude Pro | 20 / month (17 annual) | Includes Claude Code; web, desktop, mobile | Individual; 5-hour and weekly limits |
| Claude Max | 100 / month | 5× Pro usage; USD 100 API credits | Individual; check limits for heavy tasks |
| Claude Ultra | 200 / month | 10× Pro weekly limit; USD 200 API credits | Individual; not enterprise SSO/audit |
| Claude Team / Enterprise | Custom | Central admin, SSO, audit, etc. (sales quote) | No public list price obtained; do not guess |
Official list prices in USD, excluding tax, as of 31 October 2026. GitHub individual prices from Copilot plans; Enterprise from GitHub Pricing. Cursor from Cursor Pricing. Claude from Claude Pricing. Enterprise agreements, reseller pricing, tax, minimum commitments, and overage rules follow the contract and invoice.
What do 5 and 20 developers cost per month?
At public list prices, a 5-person team is roughly USD 100–150/month and a 20-person team roughly USD 400–600/month. GitHub Enterprise and Claude individual plans measure different things, so this is not the whole comparison.
The table covers seat fees only: no third-party models, overage, tax, training, or implementation. GitHub Enterprise’s USD 21 includes the repository platform and Copilot. If the company already pays for GitHub Enterprise, the incremental Copilot cost can be treated as zero; that does not make the entire Enterprise platform free.
| Plan | 5 users / month | 20 users / month | 5 users / year | 20 users / year | Basis |
|---|---|---|---|---|---|
| GitHub Enterprise (includes Copilot) | USD 105 | USD 420 | USD 1,260 | USD 5,040 | Repository platform, SSO, audit; not pure Copilot |
| GitHub Copilot Pro (individual) | USD 50 | USD 200 | USD 600 | USD 2,400 | No org SSO/audit |
| Cursor Teams (entry tier) | USD 150 | USD 600 | USD 1,800 | USD 7,200 | Excludes third-party model usage |
| Claude Pro (monthly) | USD 100 | USD 400 | USD 1,200 | USD 4,800 | Individual plans; no central admin |
| Claude mix (5 Max + 15 Pro) | — | USD 800 | — | USD 9,600 | Assumes 25% heavy users |
Editorial model: seat price × users × 12. Excludes tax, third-party models, overage, training, administration, and rework. Claude Pro annual billing at USD 17/month makes 20 seats USD 4,080/year; this table uses USD 20 monthly for comparability. Cursor Teams “30+” means higher tiers may exist; the model uses the USD 30 entry price.
Why is “USD 30 per seat” not TCO?
The second bill sits in model usage, AI Credits, third-party APIs, administrator time, security/compliance, and rework. Seat fees are only the fixed part.
The three vendors meter differently. GitHub individual plans include a pool of AI Credits; agent mode, code review, and CLI consume credits, and extra credits are purchased separately. Cursor bills third-party models at API list price, with pooled team usage and top-ups. Claude plans have 5-hour and weekly limits; Max/Ultra raise limits and include API credits, while central administration requires an enterprise discussion. Writing all three as “X dollars per seat” hides these differences.
| Cost line | Included in seat price? | How to meter | Common omission |
|---|---|---|---|
| Seat fee | Yes | Users × monthly price × 12 | Idle/leaver seats, minimum seat count |
| Model/usage overage | Usually not | AI Credits, third-party API, top-ups | Treating an allowance as unlimited |
| Admin, SSO, audit | Enterprise only | Admin time, identity integration, log retention | Mixing individual accounts; no offboarding |
| Training and migration | No | Onboarding, prompts/standards, pilot support | Buying tools without changing workflow |
| Review and rework | No | PR rework, tests, defect fixes | Measuring generation speed, not merge quality |
| Security and compliance | Partly | Code retention, training use, residency, DPA | Using personal accounts for client code |
| Exit and export | No | Code/config/history migration, termination | No export or exit terms |
Editorial guidance, not an accounting taxonomy. Follow corporate finance policy, vendor contracts, and actual invoices.
How do you rebuild the annual cost for 20 developers?
Start with fixed seat cost, then replace usage assumptions with actuals. Separate “platform already paid for” from “new AI spend”, or you will label the whole platform as AI cost.
| Scenario | Annual seat cost | Usage assumption | Annual cash cost | Fit / caveat |
|---|---|---|---|---|
| Existing GitHub Enterprise, add Copilot | USD 5,040 (full) | 0 | Incremental can be 0 | Platform already paid; do not call the full platform free |
| GitHub Copilot Pro, 20 seats | USD 2,400 | 0 | USD 2,400 | Individual; no org SSO/audit |
| Cursor Teams, 20 seats | USD 7,200 | Third-party USD 10/user/month = USD 2,400 | USD 9,600 | Assumption; set a spending cap |
| Claude Pro, 20 seats | USD 4,800 | 0 | USD 4,800 | Individual; no central admin |
| Claude mix, 5 Max + 15 Pro | USD 9,600 | 0 | USD 9,600 | Assumes 5 heavy users |
Formula: annual cash cost = 12 × seats × monthly seat price + 12 × per-user usage + one-off training/admin/compliance. Usage, heavy-user share, FX, and tax are assumptions and must be replaced with the company’s invoice. These are not quotes.
Four decision paths
Start with the existing code platform and governance, then the developers’ workflow. Do not pick a model first and bolt on identity, audit, and exit later.
| Scenario | Start with | Why | Do not ignore |
|---|---|---|---|
| Already on GitHub Enterprise with mature PR/permission/audit | GitHub Copilot | Copilot is included in Enterprise; shortest governance and billing path | AI Credit overage, code retention settings |
| IDE-first, needs multiple models/cloud agents/review | Cursor Teams | Team management, SSO, usage analytics; flexible third-party models | Third-party API bill, data retention, spending cap |
| Terminal-heavy, long tasks, agent workflows | Claude Code (Pro/Max) | Claude Code is included in the plan; higher tiers for long tasks | 5-hour/weekly limits; team admin needs Enterprise |
| Code cannot leave the intranet; strict compliance/residency | Private gateway/self-hosted or enterprise contract | Data boundary, audit, and permissions under control | Do not bypass with personal plans; require DPA/security review |
Editorial guidance, not a vendor ranking. Existing platform, language stack, network, compliance, and budget differ; run a 4–8 week limited pilot before expanding seats.
What should procurement and acceptance measure?
Measure active seats, merge quality, and cost per successful task — not generated lines or the headline seat price.
| Acceptance item | Suggested definition | Evidence |
|---|---|---|
| Active seats | Weekly/monthly active ÷ purchased seats | Admin console, login/usage logs |
| Adoption and merge | Accepted suggestions, merged PRs, reverts | Code platform records, review notes |
| Review rework | Rework rounds, test additions, defect fixes | PR history, defect tickets |
| Cost per merged PR | (Seats + usage + admin) ÷ merged PRs | Invoice, usage, merge records |
| Usage overage | AI Credits/third-party API overage | Billing detail, spending-cap records |
| Security incidents | Secrets/client code/sensitive-data exposure | Security alerts, audit logs |
| Data terms | Code retention, training use, residency, DPA | Contract, security whitepaper, config |
| Exit and export | Code/config/history migration after termination | Exit terms, export drill |
There is no universal productivity threshold. Compare the same team, task set, and time window. This article does not cite vendor customer-case productivity percentages or promise labour savings.
Who should read this — and who should not?
For R&D and procurement teams selecting AI coding tools, budgeting seats, or negotiating renewals. Not for a 1–2 person trial with no code-platform governance or usage data.
Read if: you are an engineering leader, platform/DevOps, procurement, finance, security, or compliance lead; you are comparing Copilot, Cursor, and Claude Code, or moving from individual to team plans; you need to separate seat, usage, and admin costs. Skip if: you want one article to declare the “best” tool; you have no code hosting/permission/audit baseline and plan to use personal accounts in production; you treat a vendor demo as ROI proof. Supports: seat comparison, 5/20-person models, TCO lines, decision paths, and acceptance metrics.
Lingyi does not resell these AI coding-assistant seats and does not claim any tool is “the best fit”. If code cannot leave the intranet and you need a private gateway, model routing, permissions, and audit, see LingHarness; for a lightweight trial of public tools, the vendor’s official plan is usually faster. To scope an AI project budget, selection, or acceptance checklist, talk to us.
See our separate notes on LLM API cost, AI project RFPs, and private-deployment TCO; this article only covers coding-assistant seats and usage.
FAQ
Why is comparing seat prices not enough?
Seat prices cover only the fixed subscription. GitHub AI Credits, Cursor third-party models, and Claude usage limits/API credits all change the actual bill; enterprise tiers add SSO, audit, residency, and administration. Compare the same headcount, usage assumptions, and governance requirements before comparing unit prices.
If we already pay for GitHub Enterprise, is Copilot extra?
GitHub’s official pricing page shows Enterprise at USD 21/user/month and includes Copilot. If the company already pays for GitHub Enterprise, the incremental Copilot seat cost can be treated as zero. If you buy Enterprise only for Copilot, include the repository platform, SSO, and audit in the total; do not call Copilot “free”. A standalone Copilot Business public quote was not obtained and must be confirmed with GitHub or a reseller.
Does Cursor Teams’ USD 30 include model usage?
Not third-party models. The official page lists Teams at USD 30+/user/month and bills third-party models separately at API list price, with team management and top-ups; pooled usage is an Enterprise feature. Set a spending cap and verify against the actual API invoice; do not treat USD 30 as unlimited usage.
Is Claude Code cheaper on Pro or API?
There is no universal answer. Pro at USD 20/month includes Claude Code but has 5-hour and weekly limits; heavy long tasks may need Max/Ultra or metered API usage. Compare total cost per completed task, including limits, retries, and human takeover — not the price per million tokens alone.
How do we estimate for 5 or 20 developers?
Start with public seat prices: for 5 users, GitHub Enterprise is about USD 105/month, Cursor Teams about USD 150, Claude Pro about USD 100; for 20 users, about USD 420, 600, and 400 respectively. Then add third-party/overage usage, admin, and training. The table is an editorial model, not a quote.
How should we accept AI coding-assistant ROI?
Compare the same team, task set, and time window: active seats, merged PRs, rework rounds, cost per merged PR, usage overage, security incidents, and data terms. Do not rely only on generated lines, completion counts, or vendor case-study productivity percentages; they do not prove your team will get the same result.
Method and limitations
This article uses only official pricing and public product pages accessible on 31 October 2026. It does not use private contracts, customer invoices, or vendor internal data. Seat prices, usage allowances, AI Credits, third-party model prices, and enterprise terms can change. The article has no actual enterprise agreement, tax, FX, training, administration, or rework data, so every annual cost is an editorial model, not a quote. OpenAI Codex/ChatGPT Business did not provide a verifiable public price during this check and is excluded. Claude Team/Enterprise, standalone GitHub Copilot Business, and Cursor Enterprise custom pricing remain unknown.
Sources and basis
Data cut-off: 31 October 2026. The following are official pricing/product pages; prices, allowances, and terms follow the original page and contract.
- GitHub, Copilot plans: github.com/features/copilot/plans. Copilot Pro/Pro+/Max prices, AI Credits, and feature scope.
- GitHub, Pricing: github.com/pricing. GitHub Enterprise at USD 21/user/month including Copilot; Team at USD 4/user/month.
- Cursor, Pricing: cursor.com/pricing. Plus/Premium/Super/Ultra, Teams at USD 30+/user/month, third-party models at API price, usage pools.
- Anthropic, Claude Pricing: claude.com/pricing. Pro at USD 20/month (USD 17 annual), Max 100, Ultra 200, Claude Code inclusion and usage limits.