Open WebUI vs LibreChat vs AnythingLLM: which to choose

Which self-hosted chat interface fits your company rollout

For a company-wide rollout, Open WebUI is the strongest fit when you want an offline-capable, Ollama-first shared workspace with SSO and SCIM provisioning. LibreChat suits teams that want an MIT-licensed, multi-provider chat on a MongoDB stack. AnythingLLM suits document-centric use, run as a Docker server for many users or as a single-user desktop app. Start with the license, because the Open WebUI License is the first filter. Since v0.6.6 (April 19, 2025), its branding clause applies to deployments above 50 users in a rolling 30-day period unless you hold an enterprise license or written permission from the copyright holder. Confirm your expected user count against that threshold before you build around the product, and read the license text in the Open WebUI License documentation with counsel.

The second filter is the user model. Open WebUI documents SSO through OIDC and LDAP, SCIM 2.0 provisioning, groups and per-resource permissions, and authenticator-app MFA. LibreChat supports OAuth2, LDAP and email login with an admin panel. AnythingLLM's multi-user permissions exist only in its Docker edition, so a desktop install cannot serve as a shared company tool. Map your identity requirements, including whether your directory must push group membership automatically, against these controls and verify them in a test environment, since feature availability can differ by edition and version.

The verdict depends on who will run the platform. Each option needs a database and supporting services beyond the app itself, and Open WebUI needs PostgreSQL, Redis and shared file storage for multiple replicas. LibreChat's reference stack includes MongoDB, Meilisearch and a RAG API service. Ask your platform team which stack they can patch, back up and monitor, and ask a partner for a written scope that covers those services. A fixed-scope engagement lets you price the deployment, integration and hardening work before committing to a product.

Which platform fits which company scenario

This table maps common company scenarios to the platform that fits best and the condition that decides it. Read each row as a starting point, then check the conditions against your user count, branding plans and deployment model before you commit.

Scenario	Best fit	Why, and the condition that matters
Offline or air-gapped use	Open WebUI	Open WebUI is built to run entirely offline and connects to Ollama and OpenAI-compatible endpoints. Confirm that every extension, model pull and update path works inside your network before you rely on it.
Ollama-first local models	Open WebUI	Ollama is its primary model path and it adds OpenAI, Anthropic and other compatible providers from one interface. AnythingLLM also runs local models through its built-in engine, so compare model handling in a test.
Rebranding for staff or customers	LibreChat or AnythingLLM	LibreChat is MIT licensed and AnythingLLM is MIT licensed, so neither carries a branding clause. Open WebUI branding can only be changed under its conditions: at most 50 end users in a rolling 30-day period, written permission, or an enterprise license.
Company-wide rollout above 50 users on Open WebUI	Open WebUI with an enterprise license, or another platform	Above 50 users in a rolling 30-day period, the branding clause applies unless you hold an enterprise license or written permission. Ask the vendor for the current terms in writing before you size the rollout.
Multi-provider cloud access	LibreChat	LibreChat's README lists Anthropic, AWS Bedrock, OpenAI, Azure OpenAI, Google, Vertex AI, Groq, Mistral and DeepSeek alongside custom endpoints. Open WebUI and AnythingLLM also support many providers, so confirm the specific models your teams need.
Document-heavy knowledge work with multiple users	Open WebUI or LibreChat, with the Docker deployment	Multi-user access with admin control and rule-based workspace permissions is documented for the AnythingLLM Docker version only. Open WebUI offers RBAC with groups and SSO through OIDC, so check your identity requirements against each.
Small single-team pilot	AnythingLLM Desktop	The desktop version is single-user without permissions, which suits a one-person evaluation. Move to the Docker version before adding other users, since permissions are not available on desktop.
Multi-replica or high-availability rollout	Open WebUI or LibreChat	Open WebUI needs PostgreSQL and Redis plus shared file storage across replicas. LibreChat needs MongoDB, Meilisearch and the RAG API. Confirm the operating effort of either before choosing.

How the Open WebUI License and MIT terms change the decision

The license question is the first filter for Open WebUI, and it turns on one date. Code merged up to and including v0.6.5 stays under BSD-3-Clause, but since v0.6.6 (April 19, 2025) Open WebUI is distributed under the Open WebUI License. That license forbids altering or removing Open WebUI branding in any deployment with more than 50 end users in a rolling 30-day period, unless the operator holds an enterprise license or the copyright holder gives written permission. Count users before you pick a version, not after the rollout. Read the exact terms in the Open WebUI License documentation and have counsel confirm how your staff, contractors and customers are counted.

Both LibreChat and AnythingLLM are MIT licensed, so neither carries a branding clause of this kind. LibreChat has been under ClickHouse ownership since ClickHouse announced its acquisition on November 4, 2025, and ClickHouse said existing deployments continue unchanged. AnythingLLM's README describes it as MIT licensed, with a Docker server and a desktop app. MIT is the more permissive base, and that matters if you plan to rebrand the interface for customers or partners. It does not remove the need for diligence. Confirm the license of any add-on, plugin or enterprise component you deploy alongside the core project.

MIT also does not settle data handling. AnythingLLM collects anonymous telemetry by default, which you can turn off with DISABLE_TELEMETRY=true or in the app's privacy settings. Set this in your deployment template, record the setting in your change log, and check it during acceptance testing. Ask LibreChat and Open WebUI for their own telemetry and outbound-connection behavior in writing, then compare it with your data-flow policy. Whichever license you choose, the decision depends on whether your use is internal, customer-facing or branded, and that answer should be documented before procurement.

How the three chat interfaces compare on identity, retrieval and agents

This table compares Open WebUI, LibreChat and AnythingLLM on identity and provisioning, document retrieval, agents, model providers and deployment methods. Read each row with your identity provider, document volume and deployment platform in mind, since several answers depend on edition or configuration and are marked for confirmation.

Capability	Open WebUI	LibreChat	AnythingLLM
Login and SSO	OIDC and LDAP documented; authenticator-app MFA and API keys included	OAuth2, LDAP and email login with an admin panel for users and settings	Multi-user with admin and role-based workspace access in the Docker edition only; confirm SSO support with the vendor
User provisioning	SCIM 2.0 user and group provisioning documented; confirm which identity providers are tested	Provisioning from directories not listed in the README; confirm with the vendor	Admin-managed user accounts; desktop edition is single-user without permissions
Access control model	Role-based access with groups and per-resource permissions	Admin panel for user and configuration management; confirm fine-grained resource permissions	Rule-based access to workspaces and documents in Docker; no permissions in desktop
Default vector store	Retrieval supports 13 vector databases, including PGVector, Qdrant and Milvus, with hybrid BM25 plus vector search and reranking	Reference Compose stack runs MongoDB, Meilisearch and a separate RAG API service	LanceDB is the default; PGVector, Chroma, Weaviate, Qdrant and Milvus are alternatives
Retrieval services to run	Pick a vector database; multi-replica setups also need shared file storage	Operate MongoDB, Meilisearch and the RAG API alongside the app	Works out of the box with the default store; confirm scale limits before a company-wide rollout
Agents and tools	Confirm the built-in agent and tool features against current documentation	Agents with MCP support and a sandboxed code interpreter listed in the README	No-code agent builder, MCP compatibility, scheduled tasks and a full developer API
Model providers	Ollama, OpenAI, Anthropic and any OpenAI-compatible endpoint from one interface	Anthropic, AWS Bedrock, OpenAI, Azure OpenAI, Google, Vertex AI, Groq, Mistral, DeepSeek and custom endpoints	Many cloud and local providers including OpenAI, Azure OpenAI and LiteLLM, plus a built-in local engine
Deployment methods	Docker, Docker Compose, Kubernetes with Helm and pip install	Docker, Kubernetes with a Helm chart and one-click cloud options	Desktop app for Mac, Windows and Linux, and a Docker server; the website chat widget is Docker only

License, editions and what is paid for each platform

Open WebUI moved from BSD-3-Clause to the Open WebUI License with v0.6.6, effective April 19, 2025. Code merged through v0.6.5 remains BSD-3-Clause, so an older fork stays under the permissive terms. The newer license forbids altering or removing Open WebUI branding in deployments above 50 end users in any rolling 30-day period, unless the operator holds an enterprise license or written permission from the copyright holder. Below that threshold, internal use of the platform is free of the branding clause.

Branding threshold: the 50-user limit counts end users in a rolling 30-day window, so growth past it can trigger the clause without any change to the software.
Enterprise license: the Open WebUI Enterprise documentation describes licenses that allow white-labeling, custom branding and dedicated support options.
Pricing model: pricing is not published. The vendor quotes after contact with enterprise sales and an expected seat count.
Recurring cost: the enterprise license is a commercial agreement, so ask whether it is annual, per seat or tiered, and what renewal terms apply.
Your own labor: upgrades, backups and hardening remain in your budget unless you buy a separate support arrangement.

Confirm with the vendor whether your expected user count and any customer-facing use would trigger the clause, and get the enterprise terms in writing before rollout.

AnythingLLM is MIT licensed, according to its README, and its releases include a Docker server edition and a desktop application. The Docker edition supports many simultaneous users with user management, administrator roles and rule-based access to workspaces and documents. The desktop edition is single-user and has no permissions model, so it is not a company-wide option. AnythingLLM v1.18.0 was published on October 13, 2026.

Edition split: choose the Docker server for multi-user rollouts. The desktop app suits individual use only.
Telemetry by default: the README says anonymous telemetry is on by default. Disable it with DISABLE_TELEMETRY=true in the environment or in the privacy settings of the app.
Paid options: the supplied facts describe no paid tier, so ask the maintainers whether any commercial support or hosted offering exists.
Mobile app: the README lists an open-source mobile app for on-device use, which is separate from the server rollout.
Your own labor: upgrades, backups and access reviews fall on your team unless you contract them out.

Confirm the telemetry setting in your deployment configuration and document it for your privacy review before users onboard.

What drives the running cost of each chat platform

Running any of these chat platforms costs more than the software itself. The main drivers are the supporting services each one needs, the license terms that decide whether you pay for an enterprise agreement, the staff time spent on upgrades, and the model API bills that scale with use. Weigh these together before you compare sticker terms.

Supporting infrastructure per platform

Open WebUI replicas need PostgreSQL, Redis for sessions and WebSockets, and shared file storage. LibreChat runs MongoDB, Meilisearch and a separate RAG API service. AnythingLLM ships with LanceDB by default, so its footprint is smaller until you add a client-server vector store. Ask each vendor or your own platform team to size these services in your environment, because they are the largest recurring infrastructure line.

Enterprise license and quote exposure

Open WebUI's branding clause applies above 50 users in any rolling 30-day period unless you hold an enterprise license or written permission. LibreChat (MIT) and AnythingLLM (MIT) carry no comparable branding fee. No prices are published for Open WebUI enterprise licenses, so request a written quote with your expected seat count and confirm what support is included.

Staff time for upgrades

Each project ships on its own cadence. Open WebUI v0.12.3 was published on October 16, 2026, LibreChat's latest tagged release v0.8.8 is a pre-release, and AnythingLLM v1.18.0 was published on October 13, 2026. Budget recurring engineering time for testing, migrations and rollback plans. Pre-release tags need a longer test window before production upgrades.

Model API spend

Whichever front end you choose, the largest variable cost is often token usage on cloud models such as OpenAI, Anthropic or Azure OpenAI. Local models through Ollama or LiteLLM shift that cost to GPU hardware and power. Ask for per-user usage caps, and check whether the platform gives you per-team budgets or whether you must build them in a gateway.

Backup, monitoring and on-call effort

Multi-service stacks need backup and restore plans for every stateful component, not only the chat database. Open WebUI replicas also need shared storage and Redis to stay available. Define who owns backups, restore tests and monitoring for each service, and confirm the recovery steps with the vendor before the rollout. Services you do not monitor tend to fail during peak usage.

Support and implementation scope

Implementation work is a one-time cost, while upgrades, patches and user support recur. DeployOpen sets up, hardens and supports these platforms on a fixed-scope basis, and ongoing support is available as a retainer. Ask any implementation partner for a written scope that names the infrastructure, the integrations, the upgrade cadence and the support hours before you compare totals.

How the data stores and services differ between the three

The data layer decides how much platform work sits behind each chat front end, so check it before you check features. Open WebUI runs on SQLite in a single container, but its scaling guide requires PostgreSQL instead of SQLite for multi-instance deployments, because SQLite does not handle concurrent writes from multiple processes well. Once you run more than one replica, Open WebUI also needs Redis as a shared store for sessions, websocket connections and application state, plus shared file storage so uploads and knowledge stay reachable from every replica. Confirm the current thresholds in the Scaling Open WebUI guide before you size the rollout.

LibreChat has the heaviest reference stack of the three. Its docker-compose file runs the app with MongoDB as the primary database, Meilisearch for search and a separate RAG API service for document retrieval. Each of these is a container you must back up, patch and monitor. If Meilisearch or the RAG API is missing or unhealthy, the core chat may still load while search or file-based answers fail, so test those paths directly after install. A team without a platform engineer should plan for someone who can operate MongoDB backups and restores, not just the chat container.

AnythingLLM in its Docker edition uses LanceDB as the default vector database, which is embedded and needs no separate server, so a single-container install works without extra data services. Client-server stores such as PGVector, Chroma, Weaviate, Qdrant and Milvus are optional alternatives you switch to when you need them. The practical risk is the reverse of the other two: a team that starts on the default and later wants a shared, backed-up vector store will need a migration plan. For a team without a platform engineer, the simplest path is the default for a pilot, with a written decision on when the vector store moves to a server-based option.

How a company-wide rollout of one of these platforms proceeds

The sequence moves from license and user count to a running, backed-up, staff-ready chat service, with each phase producing a check the team can sign off before the next begins.

Confirm license fit and user count
Before any install, the platform owner, legal or procurement lead and the sponsoring CIO or CTO agree on the expected number of users and how staff, contractors and customers will reach the service. For Open WebUI, this is the step where the branding clause matters: deployments above 50 users in a rolling 30-day period need an enterprise license if the branding must stay altered or removed. LibreChat (MIT) and AnythingLLM (MIT) carry no comparable clause, so the check here is mainly to confirm the license texts against your own use. Done means a signed record of user count, use type (internal or customer-facing) and the license position for the chosen platform.
Choose the deployment method
The platform team selects Docker, Docker Compose, Kubernetes with Helm or pip, and this choice sets the rest of the plan. Open WebUI lists all four install paths, and its multi-instance guidance points toward PostgreSQL and Redis. LibreChat documents Docker and a Helm chart, and its reference Compose file starts MongoDB, Meilisearch and a RAG API service. AnythingLLM's multi-user model applies only to its Docker edition, so the desktop app cannot serve a company rollout. Done means a written architecture decision that names the hosting environment, the services each platform needs and the people who will operate them.
Wire identity and user provisioning
The identity team connects the platform to the company directory and sets the role model before broad access opens. Open WebUI documents OIDC, LDAP, SCIM 2.0 provisioning and group-based role-based access control. LibreChat supports OAuth2, LDAP and email login with an admin panel. AnythingLLM's Docker edition provides user management and rule-based workspace access. Confirm group mapping, admin roles and whether deprovisioning removes access as intended. Done means a test group of users can sign in through the company identity provider, and a user removed from the directory loses access in a test run.
Select the vector store and model providers
The team picks the model providers and the vector database that will hold knowledge content. Open WebUI supports Ollama and OpenAI-compatible endpoints and lists 13 vector databases for retrieval, including PGVector, Qdrant and Milvus. LibreChat lists Anthropic, AWS Bedrock, OpenAI, Azure OpenAI, Google and Vertex AI alongside custom endpoints. AnythingLLM defaults to LanceDB and supports PGVector, Chroma, Weaviate, Qdrant and Milvus. Ask each vendor which providers and stores your data residency rules allow, and record the decision. Done means a documented provider list, a chosen vector store and a test retrieval query returning expected passages from a sample document set.
Load test, back up and restore
Before users arrive, the operations team runs a load test at the expected concurrency and rehearses a backup and restore of the database, file storage and configuration. Open WebUI's scaling guide makes PostgreSQL, Redis and shared file storage prerequisites for multiple replicas, so these components need their own restore test. LibreChat and AnythingLLM depend on their own stores, which must be included in the same routine. The sizing figures come from your own test, not from vendor claims. Done means a load test report at expected peak, and a restore that brings the service back on a clean host with users, chats and uploaded knowledge intact.
Onboard staff and set usage rules
The rollout lead publishes acceptable-use guidance, names the owners of shared knowledge bases and gives a short training session on the chat interface, workspace or group permissions and the data classes that may be uploaded. Keep the guidance platform-specific: Open WebUI knowledge bases, LibreChat agents and AnythingLLM workspaces each handle sharing differently. Configure telemetry settings before launch, since AnythingLLM collects anonymous telemetry by default and can disable it with DISABLE_TELEMETRY=true. Done means a pilot group has used the service for a set period you define, feedback has been logged, and the usage rules are signed off by security and compliance.
Run upgrade and maintenance routines
Ongoing ownership begins once users depend on the service. The platform team tracks each project's release feed and applies upgrades in a staging copy first, then runs the backup and restore test again. Open WebUI publishes versions such as v0.12.3 on its GitHub releases page, LibreChat's latest tag is a pre-release, and AnythingLLM publishes v1.18.0, so the team should read release notes before each change. Assign a named owner for patching, certificate renewal and access reviews. Done means a written maintenance calendar, a tested upgrade in staging, and a recorded rollback path for each release that reaches production.

What implementing each platform involves in your infrastructure

The implementation work differs most in the identity and data layers, so start there. For Open WebUI, a company-wide rollout means configuring OIDC or LDAP for sign-in, SCIM 2.0 for user and group provisioning, and then mapping identity provider groups to Open WebUI roles and per-resource permissions. Multi-instance operation also means moving off SQLite to PostgreSQL, adding Redis for sessions and WebSocket state, and providing shared file storage so uploads and knowledge stay reachable from every replica, as the scaling guide describes. For LibreChat, the reference stack adds MongoDB, Meilisearch and a separate RAG API service, and OAuth2, LDAP or email login with its admin panel covers user management. For AnythingLLM, the Docker edition provides user management and rule-based workspace access, while the desktop edition has no permissions, so a company rollout should use the server edition.

Each platform also needs deliberate choices about the vector store, backups and ingestion. Open WebUI supports 13 vector databases, so confirm which one your team will operate. LibreChat's RAG path depends on its bundled services, so back up MongoDB and Meilisearch together and test a restore. AnythingLLM ships with LanceDB as its default vector store, which is simple to start with but should be weighed against a client-server option such as PGVector or Qdrant for shared, long-lived data. Write a document ingestion policy before launch covering who may upload, which file types are allowed and how deletions propagate to the index. Store provider API keys as secrets rather than in user settings, and for AnythingLLM decide on telemetry early: it is on by default and can be turned off with DISABLE_TELEMETRY=true or in the privacy settings.

Effort rises with user count, identity complexity and the team's container and Kubernetes skills. Ask any implementation partner, DeployOpen included, to scope these items in writing: identity integration and group mapping, database and vector store operation, backup and restore testing, ingestion policy, provider credential handling, telemetry settings and upgrade testing against each release. Confirm with each vendor which Helm chart or Compose file it supports for your target environment, and whether your staff can run the Kubernetes components the chosen design requires. Upgrade testing deserves its own budget line, because all three projects publish frequent releases and a chat front end that holds shared knowledge and user data needs staging validation before each production change.

Questions buyers ask before choosing or hiring help

Which of these three chat interfaces is cheapest to run?

None of the three charges a software license fee for the self-hosted code, so running cost comes from the services each one needs and the people who operate them. LibreChat's reference Docker Compose file runs the app with MongoDB, Meilisearch and a separate RAG API service. Open WebUI is simplest on a single instance, but multiple replicas need PostgreSQL, Redis and shared file storage. AnythingLLM's default vector store is LanceDB, which adds no separate server. The one license cost to check is Open WebUI's branding clause, where an enterprise license is quoted after contacting sales with an expected seat count and is not published. Ask each vendor or partner for a quote that lists the services, hosting, upgrades and support as separate lines, so you can compare recurring costs directly.

Does the Open WebUI branding clause apply to internal staff?

The clause depends on how many end users the deployment has, not on whether they are employees or customers. Under the Open WebUI License, altering or removing branding is restricted in deployments above 50 end users in any rolling 30-day period, unless the operator holds an enterprise license or written permission from the copyright holder. An internal rollout to more than 50 staff members therefore falls under the restriction, and the count should include every person who logs in during that window. Smaller teams can rebrand freely within the stated limit. Read the Open WebUI License documentation with counsel, and confirm how your user count is measured before you plan a logo change, a custom name or a single sign-on login page that displays branding.

How long does a company-wide chat rollout take, and what drives it?

The duration depends on scope, not on the software alone, and no supplied fact gives a timeline for these platforms, so treat any figure from a vendor as a claim to verify. The main drivers are the identity integration, since OIDC, LDAP, SCIM or OAuth2 each need directory work and access reviews, the number of model providers and their data handling approvals, the document sources that feed retrieval, and the hosting architecture. A single-instance pilot moves faster than a multi-replica deployment that needs PostgreSQL, Redis and shared storage. Compliance evidence, user training and acceptance testing add time. Ask for a written scope with deliverables and milestones before agreeing dates, because DeployOpen works on fixed-scope engagements agreed in writing before work starts.

Can our team run this without outside help?

A team with container and identity experience can run a pilot with the official install methods, which are Docker and Docker Compose, Kubernetes with Helm, and pip install with open-webui serve. The harder part is operating the company-wide version over time. Open WebUI needs PostgreSQL rather than SQLite for multi-instance deployments, along with Redis and shared file storage. LibreChat requires MongoDB, Meilisearch and the RAG API to be kept healthy and upgraded together. AnythingLLM's Docker edition manages users and workspace rules, but its desktop edition is single-user. Assess whether you have people for backups, upgrades, patch testing and on-call coverage. DeployOpen implements, hardens, backs up, upgrades and supports self-hosted platforms in your own infrastructure, and retainers are available if you prefer ongoing help over a one-time build.

What does the rollout need from our IT staff?

Expect IT staff to own the infrastructure, identity and data decisions, even when an outside team does the build. The work typically covers the hosting platform, database and cache services, backups and restore testing, TLS and secrets handling, network egress rules for model providers, and the identity provider integration. Open WebUI's scaling guide describes what a multi-replica setup needs, including shared secret handling and file storage reachable from every replica, so review that guide early. For LibreChat, plan for MongoDB and Meilisearch operations. For AnythingLLM, decide whether telemetry should be disabled with the DISABLE_TELEMETRY setting. Ask any implementation partner which of these tasks they perform, which ones stay with your team, and how access to your environment is granted and logged.

Does ClickHouse owning LibreChat change support or the roadmap?

ClickHouse announced on November 4, 2025 that it had acquired LibreChat and said existing deployments continue unchanged, with ongoing investment in the open-source project. That statement describes the project's direction, but it does not promise a support contract, a service level or any particular feature. LibreChat remains MIT licensed, so your existing deployment does not depend on ClickHouse for its license. Check the latest tagged release before you pin a version, because v0.8.8 was marked as a pre-release. Review the ClickHouse announcement yourself, and ask any vendor or partner to state in writing what support, security fixes and upgrade paths they cover for the version you deploy.

Is AnythingLLM's multi-user model enough for a company?

It can be enough for a defined group, but only in the Docker edition. The AnythingLLM docs say the Docker version supports many simultaneous users, with user management, admin roles and rule-based access to workspaces and documents. The desktop version is single-user and has no permissions, so it does not fit a shared rollout. Compare that model with your requirements: whether you need SSO through OIDC or LDAP, SCIM provisioning, per-group permissions on knowledge bases, and audit evidence for access changes. Open WebUI documents SSO, SCIM 2.0 and per-resource permissions, and LibreChat documents OAuth2, LDAP and an admin panel. Ask AnythingLLM which of these you need in writing before you choose it, and confirm how its access rules map to your existing groups. Read the AnythingLLM Docker overview for the current model.

How hard is it to switch chat platforms later?

Switching is easier when the vector store and document pipeline are outside the chat application, and harder when chat history, user records and permissions are tied to one stack. Open WebUI and AnythingLLM both list PGVector, Qdrant and Milvus among their supported vector options, so embeddings can sometimes move to a shared store. LibreChat stores conversations in MongoDB, so a move depends on what export and import support your version provides. Ask for an export test during the pilot, record which user, group and workspace permissions must be recreated by hand, and confirm how model provider keys are handled. Plan the migration as a fixed scope with a rollback step, and budget for re-testing identity and retrieval results after the move.

How DeployOpen helps choose and implement a private chat platform

DeployOpen plans, implements, customizes, integrates, hardens, documents and supports self-hosted open-source platforms inside your own infrastructure, so the choice between Open WebUI, LibreChat and AnythingLLM can be tested against your real identity, retrieval and model requirements. We start with a written scope, then build and hand over the platform you choose.

How we scope

Confirm which user counts, internal or customer-facing use and branding needs apply, since they decide the Open WebUI License position.
Map identity needs (OIDC, LDAP, SCIM or OAuth2) and the admin roles each team requires before choosing a platform.
Agree in writing which model providers, local runtimes and document sources the rollout must cover, with timelines and deliverables.

How we build

Deploy the chosen platform in your cloud account, data center or air-gapped network, with the supporting databases and storage it requires.
Test document retrieval, permissions and provider connections against the scope, then harden configuration and access controls.
Integrate identity, logging and backup, and document each configuration choice so your team can review and reproduce it.

How we hand over

Deliver runbooks for upgrades, backups and restores, scaling changes and the license conditions that apply to your user count.
Walk your administrators through user management, workspace or group permissions and the procedure for adding model providers.
Provide evidence and configuration records that your internal audit or compliance assessors can review, noting that DeployOpen does not issue audit reports.

How we support

Ongoing support, upgrades and continuous testing are available as retainers, scoped separately from the initial fixed-scope engagement.
Retainers can cover version upgrades for the chat platform and its databases, after your team reviews release notes and licensing changes.
Ask any provider for a written quote that lists exactly what is included, who is responsible for each component and what recurs.

Open WebUI vs LibreChat vs AnythingLLM