AI Coding Assistant Cost at Enterprise Scale: Per Seat, Per Token, or Per Server?
How per-seat and usage-based AI coding tools are priced at enterprise headcount, a table of published list prices with sources checked 2026-11-05, an illustrative 30-developer scenario, and when a perpetual per-server license with unlimited developers on it, within its capacity, fits better.
This guide is written for the person who owns the budget line, not the person who chooses the editor. It does not say which is cheaper — that depends on your headcount, hardware and usage — and every price was read on the official pricing page on 2026-09-18.
How are AI coding assistants priced today?
Every major enterprise plan we checked is priced per user, and each pairs the seat with an allowance or a meter of some form. None of the vendors below publishes a per-server or unlimited-developer plan. That is the shape of the market, and it serves a great many teams well.
GitHub moved Copilot to usage-based billing on 2026-06-01: each Business or Enterprise seat includes a monthly pool of AI credits, consumed at published model rates, and plan prices did not change (checked 2026-11-05). Cursor, now part of SpaceX, prices Teams per seat, with a weekly included pool and every other use, including every third-party model, paid from a prepaid team balance (checked 2026-11-05).
Anthropic prices Claude Team per seat with Claude Code included, and publishes Enterprise as a seat price plus usage at API rates (checked 2026-11-05). Tabnine priced its platform per user on an annual subscription, with model tokens metered separately when Tabnine supplied the LLM, until Tricentis acquired it in July 2026; its pricing page now leads to a Tricentis contact form (checked 2026-11-05).
What drives the bill at enterprise headcount?
Four things, and each is documented by the vendors themselves.
Seats. The count tracks hiring and churn. GitHub’s own billing page shows the asymmetry: adding Copilot licenses mid-cycle grows the organization’s credit pool immediately, while removing them does not shrink it until the next billing cycle (GitHub Docs, checked 2026-11-05).
Allowances that reset. GitHub’s included AI credits do not carry over between months. Cursor’s included usage is allocated per user, does not transfer between team members, and refreshes weekly (Cursor Docs, checked 2026-11-05).
Usage past the allowance. GitHub enables additional usage by default for organizations and enterprises unless an administrator disables it. On current Cursor Teams plans, usage past the weekly pool, and every third-party model from the first request, is paid from a prepaid team balance at public list API prices plus Cursor’s documented Token Rate of $0.25 per million tokens on eligible third-party requests; teams still on Cursor’s legacy plans keep on-demand usage, on by default and billed in arrears. Billing details vary by plan — check the current terms. Anthropic bills extra Team and Enterprise usage at standard API rates once an organization owner enables it (all checked 2026-11-05).
Renewal. Seat count, allowance and price list are all revisited at term end. All three vendors publish spend controls — GitHub budgets set in USD, Cursor monthly team-wide limits, Anthropic organization spend limits — so each item is a legitimate control. Together they are a forecasting job that grows with the organization.
Published list prices, checked 2026-11-05
Every figure below is a published list price, read on the vendor’s official page on 2026-11-05; prices change, so follow the source before you budget. Tiers marked “not published” are sold by quote.
| Plan | Published list price | Usage beyond the seat | Source |
|---|---|---|---|
| GitHub Copilot Business | $19 per granted seat / month; 1,900 AI credits per user / month included | Additional usage at published token rates (1 credit = $0.01); enabled by default for organizations unless disabled | GitHub Docs checked 2026-11-05 |
| GitHub Copilot Enterprise | $39 per granted seat / month; 3,900 AI credits; designed for GitHub Enterprise Cloud, licensed separately | Same as Business | GitHub Docs checked 2026-11-05 |
| Cursor Teams — Premium seat | $30 per user / month; no annual Teams price published | A weekly included pool; usage past it, and every third-party model from the first request, paid from a prepaid team balance at public list API prices plus Cursor’s documented Token Rate | Cursor Docs checked 2026-11-05 |
| Cursor Teams — Super seat | $100 per user / month | Same as Premium, with a larger weekly pool | Cursor Docs checked 2026-11-05 |
| Cursor Teams — Ultra seat | $200 per user / month | Same as Premium, with the largest weekly pool | Cursor Docs checked 2026-11-05 |
| Cursor Enterprise | Not published — by quote | Pooled usage shared across the team | cursor.com/pricing checked 2026-11-05 |
| Claude Team — Standard seat (Claude Code included) | $20 per seat / month billed annually; $25 monthly; plan sold for 2–150 seats | Extra usage at standard API rates once enabled by an organization owner | claude.com/pricing · Anthropic support checked 2026-11-05 |
| Claude Team — Premium seat | $100 per seat / month billed annually; $125 monthly; plan sold for 2–150 seats | Same as Standard | claude.com/pricing checked 2026-11-05 |
| Claude Enterprise | $20 per seat / month billed annually, 20-seat minimum, plus usage at API rates | Usage cost scales with model and task | claude.com/pricing checked 2026-11-05 |
| Tabnine (acquired by Tricentis, July 2026) | No longer published: tabnine.com/pricing now leads to a Tricentis contact form. Last published price list: Code Assistant $39 and Agentic Platform $59 per user / month, annual subscription | Was: provider prices + 5% handling fee when Tabnine supplied the LLM; unlimited with your own | Archived price list, 2026-09-18 checked 2026-11-05 |
| NeueCode | One perpetual license per active server — list price shared on request; unlimited developers on it, within the server’s sized capacity; annual software assurance from year two | No per-token license fees from NeueCode; you supply and run the GPU hardware, and optional third-party model providers bill separately | NeueCode pricing model · price sheet |
The table is a like-for-like reading of published terms, not a ranking. Which shape fits depends on headcount, hardware and where the code must stay.
Illustrative scenario: 30 developers
Take a team of 30 developers and multiply each published seat price by 30 seats and 12 months. The figures are seats only and illustrative: they exclude usage beyond the included allowance, GitHub Enterprise Cloud, hardware, deployment services and any discount a vendor may offer.
| Plan | Basis | 30 developers, seats only, per year (illustrative) |
|---|---|---|
| GitHub Copilot Business | 30 × $19 × 12 | $6,840 |
| GitHub Copilot Enterprise | 30 × $39 × 12 | $14,040 — GitHub Enterprise Cloud not included |
| Cursor Teams — Premium seat | 30 × $30 × 12 | $10,800 |
| Cursor Teams — Super seat | 30 × $100 × 12 | $36,000 |
| Cursor Teams — Ultra seat | 30 × $200 × 12 | $72,000 |
| Claude Team — Standard seat, billed annually | 30 × $20 × 12 | $7,200 — 30 seats is within the plan’s 2–150 range |
| Claude Team — Premium seat, billed annually | 30 × $100 × 12 | $36,000 |
| Claude Enterprise | 30 × $20 × 12 | $7,200 + usage at API rates — 30 seats meets the 20-seat minimum |
| NeueCode | One perpetual license per active server, bought once; assurance included in year one, annual from year two | list price shared on request — how many active servers 30 developers need is set by their concurrent workload in a sizing conversation, not by headcount |
Read the last row carefully. NeueCode’s figure is not a per-year seat total but a one-time license per server plus assurance from year two, and it excludes the GPU hardware you buy and run; the per-seat figures exclude usage past the allowance. The two shapes do not sit on one line, which is why we publish no savings figure — the pricing model page states the model plainly: one fixed price per active server.
Why the token line is the hard one to forecast
Seats are countable; tokens are not, until the month is over. BCG’s 2025 analysis of B2B software pricing describes pure usage-based pricing as offering customers maximum flexibility while being the least predictable, and cites an Andreessen Horowitz survey in which 36% of respondents considering outcome-based pricing worried about cost predictability (BCG, 2025-08-13; checked 2026-09-18).
Vendors know this, which is why each vendor above publishes spend controls. When inference runs on a vendor’s cloud, the vendor meters it and publishes the rate; Tabnine’s last published price list made the same point from the other side — provider prices plus a 5% handling fee when Tabnine supplied the LLM, unlimited when the customer ran its own (archived 2026-09-18, before Tricentis acquired Tabnine).
When the model runs on hardware you own, the per-token line is replaced by a hardware line: GPU servers you buy, power and operate, budgeted rather than metered. It is a different line, not a missing one.
What a per-server license changes — and what it does not
NeueCode — Private Autonomous Intelligence for Enterprise — is one perpetual license per active server, unlimited developers on it within the server’s sized capacity, and no per-token license fees from NeueCode. The license price is set only by the number and role of licensed servers — never by a server's size: each active physical or virtual server needs its own license, bought once, and a passive disaster-recovery server is licensed at a reduced rate. Developer count, concurrent agents, repositories and model choice size the hardware — never the price.
Software assurance is included in year one and runs annually from year two as a fixed share of the license, and you can start with a free 30-day self-install trial. For a budget owner that means a license line per active server that does not move with headcount, no usage-based line items on the license, and a renewal stated as a share of the license rather than a re-count of seats.
What it does not change: you still buy and run the GPU hardware, including its electricity; a server has a capacity, and enough concurrent load calls for another licensed server; deployment services are priced separately; and optional third-party model providers, off by default, are billed by that provider if you enable one. The per-server figure comes from the price sheet — list price shared on request — and the pricing model page sets out the terms in full.

Which model fits which organization?
Per-seat pricing fits small and mid-sized teams, teams without GPU hardware, and organizations that have standardized on a cloud IDE assistant and are content with its terms. The monthly number is small, the controls are good, and there is nothing to run.
A per-server license fits large engineering organizations where seat counts move and usage is hard to forecast; regulated enterprises — government, banking, energy, healthcare, telecom and government contractors — that run or plan to run GPU hardware inside their perimeter; and programmes where the model, the prompts and the code must stay inside the network with evidence to show it. It also fits budget-governed teams that need a license line item that does not move with headcount.
Many organizations will run both: a cloud assistant where cloud is approved, and a private, on-premises deployment where it is not. The honest comparison is not a single number but a shape — countable seats plus a metered line, or countable servers plus the hardware to run them.

How to compare the two on your own numbers
Take the published list price of the per-seat plan you would actually buy and multiply it by the seats you would actually grant, on the annual or monthly basis the vendor publishes. Add the included allowance per seat and ask your engineering leads whether agentic workloads will exceed it; if so, read the overage terms and decide whether you will cap or pay. Note when the term renews and what the vendor’s policy says about added and removed seats.
On the other side, ask for a per-server list price and a sizing conversation: how many active servers your concurrency needs, whether you want a passive DR server, what assurance costs from year two, and what deployment services and hardware will add. Put both on one page — seats plus meter beside servers plus hardware — and let your own headcount and usage decide.
We publish no savings figure because there is no honest general one. The pricing model page explains the model: one fixed price per active server, quoted on request.
Frequently asked questions
Is per-seat pricing a bad model for AI coding tools?
No. It is simple, it scales down well, and every major vendor publishes spend controls for the usage line. It becomes a forecasting job at enterprise headcount, which is a different problem from being a bad model.
What does GitHub Copilot cost for a company?
Published list price on 2026-11-05: Business $19 and Enterprise $39 per granted seat per month, each with a monthly pool of AI credits consumed at published token rates; Enterprise requires GitHub Enterprise Cloud, licensed separately. Multiply by the seats you grant; additional usage beyond the pool is enabled by default unless an administrator disables it.
Is Claude Code included in the Claude Team plan?
Yes. Anthropic’s pricing page and support documentation state that Claude Code is included with every Team seat (Standard $20 and Premium $100 per seat per month billed annually; the Team plan is sold for 2–150 seats) and with the Enterprise seat (20-seat minimum), where usage beyond the seat is billed at API rates. Checked 2026-11-05.
Why is an AI coding bill hard to predict?
Because the seat is only the countable part. Included allowances reset monthly, agentic workloads consume tokens unevenly, and past the allowance most vendors bill on demand by default. BCG describes pure usage-based pricing as offering maximum flexibility to customers while being the least predictable (2025-08-13).
Can you self-host an AI coding assistant, and what does it cost?
Yes. Tabnine offers on-premises and air-gapped deployment priced per user, with token metering that does not apply when you supply your own LLM. NeueCode runs local open-weight models on your GPUs and is priced per active server as a perpetual license with unlimited developers on it within the server’s sized capacity and no per-token license fees from NeueCode; the per-server figure is on the price sheet (list price shared on request), and you supply and run the hardware.
Does NeueCode publish a per-developer price?
No. The license is per active server and does not change with headcount, so any per-developer figure is specific to a given server count and team and changes with every hire. The unit is the server: ask for the list price and a sizing conversation, and divide by your own headcount if that is useful to you.