Overview
The Collabora Online server code is licensed under MPL-2.0, so the recurring cost of running it does not come from a server license. It comes from three places: the edition you choose, the hosting and replicas you run, and the integration and patching labor your team or a partner spends keeping it current. Start by separating those lines in your budget request, because a subscription fee is only one of them.
There are three edition choices. CODE, the free edition, is released roughly once a month as a rolling build, and Collabora describes it as suited to testing, home use or small teams but not recommended for production. Business costs 3 euros per user per month for up to 99 users and includes long term support, a service level agreement, signed security updates and customer portal access. Enterprise is quoted for 100 or more users with volume discounts and adds integration support, customization services and roadmap input. Confirm current terms on Collabora's pricing page before you model figures.
Four events usually trigger a fresh cost review: a budget request, a pilot moving toward production, user counts growing past 99, or an annual review of the platform. Each one changes the edition question, so record the user count, the expected concurrent editing peak and the production date in your cost model from the start.
Most teams model hosting and internal effort themselves, since they know their infrastructure prices and staff time. A partner usually helps size the environment, price the integration work and write a cost model that separates recurring lines from one-time work. A partner does not remove subscription fees, which still apply to Business or Enterprise for as long as you use them.
What the cost model covers
A fixed-scope cost model separates what the client pays Collabora, what it pays infrastructure and partners, and what it spends on its own staff. Each category should show whether the cost is one-time or recurring.
Edition and subscription
The client pays Collabora directly for the edition, and this cost recurs. The Collabora pricing page lists the Business plan at 3 euros per user per month for up to 99 users, with long term support, an SLA and signed security updates. Enterprise, for 100 or more users, is quoted with volume discounts. CODE is free but not recommended for production. Confirm the user count basis and current pricing before modeling.
Production and non-production infrastructure
The client pays its cloud or hardware provider for compute, storage and network, and these costs recur. The Helm chart README recommends production pods with requests of 4 cores and 6000 MiB of memory and limits of 8 cores and 8000 MiB. Autoscaling is disabled, so replica counts are set by hand. A staging environment for testing updates adds a second footprint that the model must include.
WOPI host integration
This work connects Collabora Online to the host, such as Nextcloud, and is mostly one-time setup with some recurring change. It covers the server URL, the WOPI allow list, protocol and TLS alignment, group-based edit rules and routing by WOPISrc. A partner can carry it under a written scope, or the client's own team can do it. Ask whether changes after each host version update are included in the quote.
Patch and upgrade effort
This is recurring work, performed by the client's staff or by a partner under a retainer. Collabora publishes point releases and security advisories, such as 25.04.7.5, which fixed CVE-2026-23623 and was released on January 29, 2026. The 25.04 branch was still receiving point releases in 2026. Budget time to test each point release in non-production and to plan moves between yearly branches.
Support and internal skills
Support is partly bought and partly staffed. Per the pricing page, Business includes an SLA and long term support, and Enterprise adds integration support and customization services. Internal skills in Linux, reverse proxies, TLS and Nextcloud administration still carry cost, paid through salaries or partner hours. Ask which issues the subscription covers and who answers them.
Deliverables a buyer should expect
A fixed-scope engagement should end with a written cost model that separates one-time and recurring lines. Expect the sizing assumptions behind it, such as concurrent editors, user count and pod counts, along with an edition recommendation and its reasons. The quote should be tied to the written scope, and the run budget should be detailed enough for finance and operations owners to review line by line.
What drives cost
Each driver shows why it changes cost and how to keep it down, so you can ask for a quote that matches your actual scope.
| Driver | Why it moves cost | How to keep it down |
|---|---|---|
| Named users | Business is 3 euros per user per month for up to 99 users, per Collabora's pricing page. Moving from 99 to 100 users changes the plan to Enterprise, which is quoted. | Confirm your exact named user count before buying. At 100 or more, ask for an Enterprise quote with volume discounts. |
| Peak concurrent editing | Simultaneous editors, not total users, determine how many coolwsd pods and how much CPU you need. | Measure peak simultaneous sessions from host logs, then size for that peak with headroom agreed in the quote. |
| Pods and manual replicas | Chart guidance sets production limits at 8 CPU cores and 8000 MiB memory per pod, so pod count drives infrastructure cost directly. | Keep replicas to measured need, add failover capacity on purpose, and keep requests at 4 cores and 6000 MiB unless load tests show otherwise. |
| High availability routing | Autoscaling is disabled because scaling breaks collaborative editing, and users on one document must reach the same pod. | Plan WOPISrc-based routing at the load balancer or ingress, and test failover before go-live. |
| WOPI hosts and environments | Each WOPI host, such as Nextcloud, and each environment such as test, staging and production needs its own server configuration and allow list. | Share a test server only where isolation is not required, and consolidate hosts where your platform allows. |
| Document formats and file sizes | Large or complex docx, xlsx and pptx files use more memory and conversion time, which raises pod sizing. | Test your largest real working files during the pilot, and set file size and format limits at the host. |
| Integration and customization | Enterprise adds integration support, customization services and roadmap input. Custom branding or host integration adds effort that depends on scope. | Limit custom work to branding and templates where possible, and ask for named deliverables in a fixed-scope quote. |
| Compliance and security review | Evidence of allow lists, TLS, patch cadence and access control takes review time. Business includes signed security updates. | Confirm with your assessor which evidence they expect, and collect allow list and update records early. |
| AI assistant (26.04) | Added in 26.04 across Writer, Calc and Impress and off by default. Choosing an external provider adds data flow and cost review. | Keep it off until a data review approves a provider, either hosted externally or on your own infrastructure. |
Phases of a cost model
This sequence turns user counts, peak load and the edition choice into a written cost model that a finance owner and an IT owner can both sign.
Confirm user counts and edition boundary
Count named users per WOPI host, including groups that will only view documents, and confirm which edition applies. The Business plan is listed for up to 99 users and the Enterprise plan for 100 or more, so the count decides the subscription line. Confirm current pricing with Collabora. Output: a user register and an edition decision, signed off by the IT owner and procurement.
Confirm peak concurrency and document mix
Measure or estimate the peak number of simultaneous editing sessions, not total users, and record the share of docx, xlsx, pptx, PDF and large spreadsheets. Concurrency and file types drive server count and memory. Output: a peak concurrency figure with a document mix table, signed off by the business application owner and the platform lead.
Size production and test environments per server
Translate peak concurrency into the number of coolwsd servers or Kubernetes pods, plus headroom for failover and a separate test environment. The official Helm chart publishes production resource limits of 8 CPU cores and 8000 MiB per pod, with requests of 4 cores and 6000 MiB, which is a starting point to validate, not a sizing result. Output: a sizing sheet signed off by the platform engineering lead.
Cost integration and support for the edition
Price the work to connect Collabora Online to the WOPI host, including ingress, TLS, allow lists and group rights, and compare it with the support scope of the chosen edition. Enterprise lists integration support and customization services, while Business includes long term support, an SLA and signed security updates. Output: a costed work list and support comparison, signed off by the security lead and procurement.
Produce the written cost model
Assemble one-time lines such as deployment, integration, testing, data checks and staff training, and recurring lines such as subscription seats, infrastructure, backups and time spent on point releases. For each line, record what drives it up or down and what to request in a quote. Output: a written cost model, signed off by the finance owner.
Hand over the run budget and review points
Convert the recurring lines into an annual run budget and set review triggers, such as user counts approaching the Business limit, changes in peak concurrency, a new WOPI host, or a new yearly release branch. Agree the review cadence in the written scope. Output: a run budget and review schedule, approved by the CIO or budget holder.
Where cost models go wrong
These are the errors that most often turn a low estimate into an unplanned bill. Each card names the cost that gets missed and the check to run before the budget is approved.
Running CODE in production
CODE is free and suited to testing, home use or small teams, and Collabora does not recommend it for production. Teams that run it anyway often pay later, when a document outage or a security fix must be handled without a support agreement. Before go-live, confirm whether the budget includes a Business or Enterprise subscription and get the support terms in writing.
Counting only active editors
The Business plan is priced at 3 euros per user per month for up to 99 users, so the licence count follows the users covered, which can be larger than the number editing at peak. Confirm with Collabora how users are counted, then size the quote from your full eligible user base in the host platform.
Replica count drives hosting cost
The Helm chart disables horizontal pod autoscaling because scaling breaks collaborative editing and copy and paste, so replicas are set by hand and hosting cost follows that number. The chart's production figures are 4 cores and 6000 MiB requested per pod, with limits of 8 cores and 8000 MiB. Budget for at least two replicas with WOPISrc routing, plus peak capacity, rather than for autoscaling savings.
Staging and test environments
Upgrades and integration changes need a non-production copy of the WOPI host and Collabora Online, and that copy uses compute, storage and possibly licence seats. Teams often budget only the production cluster. Include staging in the quote and confirm with Collabora whether the subscription covers non-production instances.
Patch and upgrade effort
Collabora ships point releases on each yearly branch. The 25.04 branch first shipped on June 19, 2025 as 25.04.2.3 and was still receiving point releases, with 25.04.13 released on September 21, 2026. The 25.04.7.5 security fix, released January 29, 2026, also updated package signing keys for RPM and DEB users. Budget staff time to test and roll out each point release, not only the first deployment.
Unpriced AI provider usage
The optional AI assistant introduced in Collabora Online 26.04 is off by default. When you enable it, administrators choose the AI provider and whether it runs externally or on your own infrastructure. External providers bill under their own terms, so ask for the rate model and set usage limits before enabling the assistant for any group.
Recurring costs after go-live
This table lists each recurring and one-time cost of running Collabora Online, what drives it, and what to request in a quote so that CODE and subscription options can be compared on the same basis.
| Cost component | One-time or recurring | What drives it | What to ask for in a quote |
|---|---|---|---|
| Edition subscription or CODE effort | Recurring | CODE is free but not recommended for production. Business is listed at 3 euros per user per month for up to 99 users, per the Collabora pricing page. | Confirm the current listed price, the user count basis, and what a 100 or more user Enterprise quote includes. |
| Production and non-production hosting | Recurring | Helm chart README production limits are 8 CPU cores and 8000 MiB per pod, with requests of 4 cores and 6000 MiB. Staging doubles this if mirrored. | Cores and memory requested per environment, and whether non-production runs the same pod sizes as production. |
| Replica count | Recurring | Horizontal autoscaling is disabled, so replicas are set by hand. Peak concurrent editing per document and per server sets the count. | Sizing based on your peak concurrent editors, plus a spare replica for node failure and maintenance windows. |
| Backups and monitoring | Recurring | Storage and tooling for configuration, WOPI host data and server health. Retention length and restore test frequency change the volume. | Backup scope, retention period, restore test schedule, and who receives monitoring alerts. |
| Patch and point release effort | Recurring | Each branch receives point releases, such as 25.04.13 on September 21, 2026, and security fixes like 25.04.7.5 for CVE-2026-23623. | Staging test window per release, who applies signed package updates, and whether security fixes are covered by the support plan. |
| Integration maintenance after WOPI host upgrades | Recurring | Host upgrades can change WOPI behavior, allow lists and server URL settings, such as those in the Nextcloud administration manual. | Compatibility check per host release, who runs it, and who fixes configuration drift after each host upgrade. |
| Support contract | Recurring | Business includes long term support, an SLA and signed security updates. Enterprise adds integration support and customization services. | Response times, SLA terms, escalation path, and whether implementation retainers are priced separately from the subscription. |
| Internal administrator skills | Recurring staff time, one-time training | Container or Kubernetes operations, reverse proxy and TLS configuration, and WOPI host administration. Skill gaps raise support spend. | Training plan for named staff, on-call coverage, and a handover checklist that the provider will deliver. |
| AI provider usage | Recurring | The AI assistant in 26.04 is off by default. Its cost depends on the provider you choose and whether it is hosted externally or on your infrastructure. | Confirm pricing and data terms directly with the chosen AI provider, since Collabora does not set those costs. |
How cost changes by starting point
A pilot on CODE usually has no subscription cost, but Collabora describes CODE as suited to testing, home use or small teams and not recommended for production. Moving to production changes several lines at once.
- Subscription: moves from zero to the Business plan if you stay at 99 users or fewer, priced at 3 euros per user per month according to the Collabora pricing page.
- Infrastructure: stays broadly flat in topology but usually grows in redundancy, monitoring and backups.
- Update effort: shifts from following a monthly rolling CODE build to planning around supported release branches and point releases.
- Integration effort: stays flat if the WOPI host configuration from the pilot carries over.
- Hardening work: adds one-time effort for TLS, access rules and a review of the pilot configuration.
Ask the vendor which support terms apply after the move and whether they start on the signed date.
The Business plan is priced per user for up to 99 users. From 100 users, Collabora quotes the Enterprise plan with volume discounts, so the subscription line stops being a simple per-seat calculation.
- Subscription: changes from a per-user list rate to a quote, which you should request in writing for your exact user count.
- Support scope: adds integration support, customization services and roadmap input under Enterprise, according to the Collabora pricing page.
- Infrastructure per concurrent editor: stays roughly flat, since user count alone does not set server load. Concurrent editing does.
- Integration and customization effort: may fall if you need the vendor's integration support, or rise if you add customization work.
Illustrative example: a 120 user organization compares the Enterprise quote against its own estimate of integration hours saved, not against a per-seat multiplication.
The official Helm chart shifts cost toward cluster capacity and planning. The chart README states that horizontal pod autoscaling is disabled because scaling breaks collaborative editing, so replica counts are set manually.
- Infrastructure: rises with the number of pods you reserve. The README recommends production limits of 8 CPU cores and 8000 MiB of memory per pod, with requests of 4 cores and 6000 MiB.
- Cluster platform: stays flat if you already run a cluster that can host these pods.
- Ingress and routing: requires load balancer configuration that routes by the WOPISrc parameter, which adds configuration effort.
- Operations effort: rises with manual capacity planning for peak concurrent editing.
- Subscription: depends on your user count, not on the number of pods.
Confirm with Collabora whether the Collabora Online Controller applies to your scale and whether its terms affect the subscription quote.
In a network where AI hosting and downloads must stay internal, cost moves into mirroring, local hosting and staff time. Confirm each delivery method before budgeting.
- Download mirroring: adds recurring effort to pull images or packages, verify signatures and move them across the boundary for each release.
- AI assistant hosting: if you enable the optional assistant in Collabora Online 26.04, it is off by default and the administrator chooses the provider. Hosting it on your own infrastructure adds compute and model hosting cost.
- Infrastructure: stays flat for the editing servers themselves.
- Subscription: ask whether the plan's security updates and support can be delivered without internet access.
- Update effort: rises because each point release needs a controlled transfer and a test cycle.
If the AI assistant stays disabled, the AI hosting line does not apply.
Nextcloud Office can point to the CODE server built into the deployment or to a dedicated Collabora server. The choice changes infrastructure and operations lines while the subscription depends on your user count and support needs.
- Built-in CODE: keeps separate server infrastructure low, but CODE is not recommended for production according to Collabora.
- Dedicated server: adds a separate host or container, its own TLS certificates, monitoring and update path.
- Network controls: the Nextcloud administration manual recommends restricting WOPI requests to the Collabora server IP addresses and using HTTPS for the server URL. This adds configuration effort in both cases.
- Group rules: limiting which groups can edit or only view does not change license cost.
- Subscription: applies to the dedicated production server path, so confirm which setup your quote covers.
What the client needs to provide
A cost model for Collabora Online cannot be written until the buyer supplies an exact named user census, split by the groups that will hold editing rights, with a growth expectation for each group over the next budget cycle. The census owner is usually the HR or identity lead, working with the Nextcloud or WOPI host administrator who can confirm which accounts can open documents at all. Subscription pricing is quoted per user, so a rough headcount is not enough. Ask for the count of users who will edit, separate from those who only view, because the cost model and the group rules in the host both depend on that split.
The buyer also names two decision owners: the person who holds the edition decision (free CODE or a Business or Enterprise subscription) and the person who approves the spend. Hosting budget sits with a separate owner, often infrastructure or platform engineering, who confirms which cloud accounts, data centers or Kubernetes clusters will run the coolwsd servers and who pays for them. Name each of these people in writing. The Business plan is quoted per user for up to 99 users, and the Enterprise plan is quoted separately for 100 or more, so the edition owner must agree on the user threshold before the quote is requested.
Next, collect usage data. Peak concurrent editing sessions, measured over a normal busy hour and a month-end peak, drive the number of server instances and their resource requests. If the current document platform already logs file opens or session counts, the host administrator can export them. Also record the WOPI host and its exact version, the number of environments needed (production, staging and test at minimum, with names and owners), the document formats in use and typical file sizes for docx, xlsx, pptx, PDF and any Visio files. Large spreadsheets and image-heavy presentations change memory needs more than user counts do.
Finally, the buyer decides whether the AI assistant introduced in the 26.04 release may be used in Writer, Calc and Impress, and, if so, where its provider may be hosted. The assistant is off by default and the administrator chooses an external or self-hosted provider, so the answer affects hosting scope, data handling review and possibly a separate line in the cost model. If any item is missing, the cost model will carry stated assumptions for the affected lines, and each assumption should be signed off by the named owner so that the quote can be revised when the real figure arrives.
Questions buyers ask
Is Collabora Online CODE free to run in production?
No. Collabora describes CODE, its free edition, as suited to testing, home use or small teams, and not recommended for production. CODE is a rolling release with builds published on average once a month, so teams that need a stable release line usually plan around the Business subscription, which includes long term support, a service level agreement and signed security updates. Weigh the subscription fee against the administrator time needed to keep an unsupported build current and secure.
How much does the Collabora Online Business plan cost, and how are users counted?
The Business plan costs 3 euros per user per month for up to 99 users, according to Collabora's pricing page. It includes long term support, a service level agreement, signed security updates and customer portal access. Confirm with Collabora whether seats are counted per named user or per concurrent editor, because that choice changes the monthly total and the server capacity you need. Also confirm whether created but inactive accounts count toward the 99 user limit.
When does a Collabora Online subscription pay for itself?
It pays for itself when the monthly fee is lower than the internal cost it avoids, such as administrator time spent on upgrades, patching and incidents. Illustrative calculation: at 3 euros per user, 40 users cost 120 euros per month. Compare that figure with the hours your administrators would spend keeping an unsupported build current, multiplied by your internal hourly cost, plus any hosting savings you can verify. Replace every figure with your own numbers before deciding.
What does Collabora Enterprise add and how is it quoted?
Enterprise is for 100 or more users and is quoted with volume discounts, according to Collabora's pricing page. On top of the Business inclusions, it adds integration support, customization services and roadmap input. Ask Collabora for a written quote that states the user count, the support level, the customization scope and the term. Hosting stays your cost under either plan, because Collabora Online runs in your own infrastructure.
Does the Collabora Online AI assistant add to the cost?
The AI assistant, introduced in version 26.04, is off by default, so it adds no cost until an administrator turns it on. Administrators choose the AI provider and whether it runs externally or on their own infrastructure, and that choice drives the cost. An external provider bills on its own terms, while self-hosted models add compute, storage and operations work. Confirm with Collabora whether the assistant is included in your subscription, and with your chosen provider how usage is billed.
Can our team build a cost model ourselves, and how long would it take?
Yes, your team can build the cost model alone if it has the sizing and integration skills described on this page, including sizing coolwsd servers, configuring the reverse proxy and TLS, and connecting the WOPI host. The time depends on how many servers, hosts and user groups are in scope, and the duration is set in the written scope. Ask for a plan that lists the deliverables, the assumptions behind each cost line and the dates agreed before work starts. Include recurring upgrades and point releases in the estimate.
How DeployOpen scopes and supports this work
DeployOpen treats the cost of Collabora Online as a set of separate lines: edition fees, server infrastructure, integration effort and ongoing support. We set these out in a written scope so you can compare a CODE-based deployment with a subscription on the same terms.
How we scope
- The written scope lists the editions compared, the number of users and expected concurrent editors, and the WOPI hosts involved.
- Infrastructure lines cover the coolwsd servers or Kubernetes pods, load balancing, storage and backups, each named with its owner.
- Engagements are fixed-scope, so scope, deliverables and timelines are agreed in writing before any work starts.
How we build
- We size the server tier from your concurrency assumptions and document the resource requests and limits we plan to use.
- We test document round trips for the formats you rely on, such as docx, xlsx and pptx, against the chosen edition.
- We record the assumptions behind each cost line so you can re-check them if user counts or document volumes change.
How we hand over
- You receive a cost model that separates one-time setup effort from recurring subscription, infrastructure and support lines.
- The handover pack includes the deployment record, sizing notes, the edition decision and the update procedure for your team.
- Your team can confirm the subscription terms, including support and security update coverage, with Collabora directly before signing.
How we support
- Ongoing support, upgrades and continuous testing are available as retainers, so recurring effort is budgeted explicitly.
- Point releases and security advisories are tracked against the edition you run, and the retainer scope states who applies them.
- Review the cost model at each planned upgrade or user-count change, and ask us to re-quote the scope before adding work.
