AIplane feature comparison

This is a feature-by-feature comparison of croit AIplane, LiteLLM and Open WebUI. It is maintained as a standalone product document; the README remains focused on AIplane.

Checked on 6 October 2026. AIplane entries use the local main implementation revision pinned at the end of this document. LiteLLM and Open WebUI entries describe the public product documentation linked in the source index. For LiteLLM, ✅ means available in the open-source edition; 🟡 Enterprise means the capability requires a paid Enterprise license. Product capabilities change; recheck those sources before using this matrix in a proposal or contract.

The request-priority rows were checked on 11 October 2026 against AIplane revision d5b81c2a and the official documentation cited as L31 and O6. Other entries retain the review dates and revision pins below.

How to read the matrix

Mark	Meaning
✅	The feature is directly documented and available in the product.
🟡	The feature exists with a material limitation, configuration prerequisite, external service, plugin, or edition dependency. For LiteLLM, the note identifies Enterprise-only capabilities. Read the linked evidence.
❌	The reviewed product documentation explicitly says the feature is unavailable or the product surface does not provide it. This is a scoped product comparison, not a claim that no workaround exists.
⚪	Not established by the reviewed sources. Treat this as unknown, not as a product gap.
—	Not applicable to the product's role in this row.

“Partial” is not a score. It means that the feature's scope or operating conditions differ. A green LiteLLM mark means its OSS edition; it does not include Enterprise capabilities. A green mark does not mean the implementations are equivalent in quality, scale, security posture, or support. The matrix compares documented product behavior, not marketing claims, and does not infer a feature from a protocol alone.

Product scope and model access

Feature	croit AIplane	LiteLLM	Open WebUI	Evidence
Self-hosted deployment	✅	✅	✅	A1 L1 O1
Browser-based employee workspace	✅	🟡	✅	A2 L1 O1
Centralized model gateway for applications	✅	✅	🟡	A3 L1 O1
Gateway and employee workspace in one product	✅	🟡	🟡	A2 A3 L1 O1
OpenAI-compatible client interface	✅	✅	✅	A3 L2 O1
Anthropic Messages-compatible client interface	✅	✅	🟡	A3 L2 O1
Python client SDK for application developers	⚪	✅	⚪	L3
Native model-provider adapters	🟡	✅	🟡	A4 L4 O1
OpenAI-compatible upstream backends	✅	✅	✅	A4 L4 O1
More than 100 provider integrations advertised	⚪	✅	⚪	L4
Ollama integration	🟡	✅	✅	A4 L4 O1
Hosted model services through compatible upstreams	🟡	✅	🟡	A4 L4 O1
Self-hosted inference services through compatible upstreams	✅	✅	✅	A4 L4 O1
Anthropic-native upstream support	✅	✅	🟡	A4 L4 O1
TypeSafe System One upstream support	✅	⚪	⚪	A4
OpenAI Chat Completions endpoint	✅	✅	✅	A3 L2 O1
OpenAI Responses endpoint	✅	✅	🟡	A3 L28 O24
Anthropic Messages endpoint	✅	✅	✅	A3 L2 O24
Embeddings endpoint	✅	✅	🟡	A3 L2 O24
Image generation endpoint	✅	✅	⚪	A3 L28
Image editing endpoint	✅	✅	⚪	A3 L28
Audio transcription endpoint	✅	✅	⚪	A3 L28
Text-to-speech endpoint	✅	✅	⚪	A3 L28
Batch inference endpoint	❌	✅	⚪	A3 L28
Public reranking endpoint	✅	✅	⚪	A3 L28
Public moderation endpoint	❌	✅	⚪	A3 L28
OpenAI-compatible streaming	✅	✅	✅	A3 L2 O1
Provider-specific request options pass through where supported	🟡	✅	🟡	A3 L2 O1

Gateway routing and resilience

Feature	croit AIplane	LiteLLM	Open WebUI	Evidence
Multiple configured model backends	✅	✅	✅	A5 L5 O2
Model pools or groups	✅	✅	🟡	A5 L5 O2
Pool-level backend membership	✅	✅	🟡	A5 L5 O2
Load balancing across backends	✅	✅	🟡	A5 L5 O2
Round-robin selection	✅	✅	⚪	A5 L5
Least-in-flight selection	✅	✅	⚪	A5 L5
Prompt-prefix affinity selection	✅	🟡	⚪	A5 L5
Backend health probes	✅	✅	⚪	A5 L5 O2
Per-backend in-flight limits	✅	✅	⚪	A5 L5
Priority queue for model requests	✅	🟡 Beta; testing only	⚪	A36 L31 O6
Administrator-assigned Low/Medium/High priorities on groups and API tokens	✅	⚪	⚪	A36 L31 O6
Translate trusted priorities into backend scheduler parameters	🟡 Requires backend support and enabled scheduling	⚪	⚪	A36 L31 O6
Automatic route selection by quality, balance, or cost	✅	🟡	🟡	A6 L5 O3
Stable model aliases	✅	✅	✅	A5 L5 O2
Unknown-model fallback	✅	✅	🟡	A5 L5 O2
Offline-capacity fallback	✅	✅	⚪	A5 L5
Retry another replica after eligible pre-output failures	✅	🟡	⚪	A7 L5
Wait for a known backend to recover within a configured budget	✅	⚪	⚪	A7
Retry-After response when gateway capacity is exhausted	✅	✅	⚪	A7 L5
Shadow rollout for automatic routes	✅	🟡	⚪	A6 L6
Route decision history and outcome inspection	✅	✅	⚪	A6 L6
Route session affinity	✅	✅	⚪	A6 L6
Traffic mirroring to a silent model	⚪	✅	⚪	L6
A/B model experiments	🟡	✅	✅	A6 L6 O4
Cost-aware model choice	✅	✅	🟡	A6 L5 O3
Provider/model capability catalog	✅	✅	✅	A5 L4 O2
Per-model capability overrides	✅	🟡	✅	A5 L5 O2
Global model defaults	✅	✅	✅	A5 L5 O2

Identity, access and governance

Feature	croit AIplane	LiteLLM	Open WebUI	Evidence
Multi-user access	✅	✅	✅	A8 L7 O5
OIDC single sign-on	✅	🟡 (free up to 5 users; Enterprise above that)	✅	A8 L29 O5
End-user local account/password sign-in	❌	—	✅	A8 O5
Basic username/password protection for the admin UI	⚪	✅	✅	L29 O5
LDAP authentication	⚪	⚪	✅	O5
SCIM user provisioning	⚪	🟡 Enterprise	✅	L29 O5
Identity-provider OIDC/JWT authentication	✅	🟡 Enterprise	✅	A8 L29 O5
Map identity-provider groups to product permissions	✅	⚪	✅	A8 L7 O5
Basic global proxy roles and user/team/key controls	✅	✅ OSS	✅	A8 L7 O5 L29
Organization/team-scoped delegated administrator roles	✅	🟡 Enterprise	⚪	A8 L7 L29 O5
Fine-grained team-member permissions	✅	🟡 Enterprise	⚪	A8 L7 L29 O5
Per-model access restrictions	✅	✅	✅	A8 L7 O5
Per-tool access restrictions	✅	✅	✅	A8 L7 O5
Per-agent access restrictions	✅	🟡	🟡	A9 L8 O5
Per-knowledge-collection access restrictions	✅	🟡	✅	A10 L8 O5
Personal API tokens	✅	✅	✅	A8 L7 O5
Revocable API tokens	✅	✅	✅	A8 L7 O5
Token model restrictions	✅	✅	🟡	A8 L7 O5
Token tool restrictions	✅	✅	🟡	A8 L7 O5
Token request quotas	✅	✅	⚪	A8 L9
Token token-volume quotas	✅	✅	⚪	A8 L9
Token cost quotas	✅	✅	⚪	A8 L9
Group/role usage limits	✅	✅	⚪	A8 L9
Per-user usage limits	✅	✅	⚪	A8 L9 O6
Per-agent/system-principal limits	✅	🟡	⚪	A8 L9
Global usage limits	✅	✅	⚪	A8 L9
Per-model usage limits	✅	✅ (model-level controls)	⚪	A8 L9
Different model budgets for each virtual key	⚪	🟡 Enterprise	⚪	L29
Usage reporting by user	✅	✅	✅	A8 L10 O6
Usage reporting by model	✅	✅	✅	A8 L10 O6
Usage reporting by API token	✅	✅	⚪	A8 L10
Usage reporting by request source	✅	🟡	⚪	A8 L10
Cost reporting based on configured prices	✅	✅	🟡	A8 L10 O6
Unpriced usage is visibly identified	✅	🟡	⚪	A8 L10
GDPR coverage declaration per model pool	✅	⚪	⚪	A11
NDA coverage declaration per model pool	✅	⚪	⚪	A11
Content guard can monitor, confirm, or deny a request	✅	🟡 (guardrail framework OSS; behavior varies)	⚪	A11 L11 L29
Built-in moderation callback integrations	🟡	🟡 Enterprise for listed integrations	⚪	A11 L29
Guardrails scoped to individual keys or teams	✅	🟡 Enterprise	⚪	A11 L29
Human approval before a tool action	✅	🟡	⚪	A12 L11 O7
Secure input collection during a workflow	✅	⚪	⚪	A12
Append-only agent activity records with integrity verification	✅	⚪	⚪	A13
Admin impersonation with an audit record	✅	⚪	⚪	A8

Employee workspace and collaboration

Feature	croit AIplane	LiteLLM	Open WebUI	Evidence
Persistent multi-conversation chat	✅	🟡	✅	A2 L1 O1
Streaming chat replies	✅	🟡	✅	A2 L2 O1
Edit and retry a message	✅	⚪	✅	A2 O8
Interrupt a running response	✅	⚪	✅	A2 O8
Queue a follow-up while the model is responding	✅	⚪	✅	A2 O1
Share a conversation	✅	⚪	✅	A2 O1
Fork a conversation	✅	⚪	✅	A2 O8
Export a conversation as Markdown	✅	⚪	✅	A2 O8
Export a conversation as PDF	✅	⚪	⚪	A2
Compare multiple models side by side	⚪	⚪	✅	O1
Shared team conversation channels	⚪	⚪	✅	O9
Threaded replies and reactions in shared channels	⚪	⚪	✅	O9
Chat folders, tags, and pins	🟡	⚪	✅	A2 O1
Personal memory across conversations	✅	🟡	✅	A14 L12 O1
User-editable memory and preferences	✅	⚪	✅	A14 O1
Reusable personal skills/instructions	✅	🟡	✅	A15 L13 O3
Temporary chats	⚪	⚪	✅	O10
Message autocomplete	⚪	⚪	✅	O8
Follow-up suggestions	⚪	⚪	✅	O8
Reasoning/thinking display	🟡	🟡	✅	A2 L2 O8
Multilingual interface	✅	⚪	⚪	A16
Six interface languages documented for AIplane	✅	⚪	⚪	A16
Mobile-responsive chat interface	✅	⚪	✅	A2 O1
Maintained first-party extension controls the user's logged-in browser	✅	⚪	❌	A17 O22 O23
Browser automation available through tools or extensions	✅	🟡	🟡	A17 L16 O3 O15
Browser page navigation, reading, clicking, and typing	✅	⚪	🟡	A17 O3 O15
Browser screenshot of the full page or a selected region	✅	⚪	🟡	A17 O15 O23
Browser control requires explicit user enablement	✅	⚪	🟡	A17 O3
Conversation-linked editable documents	✅	⚪	🟡	A18 O1
Document revision history and restore	✅	⚪	⚪	A18
Generated file downloads from chat	✅	⚪	✅	A18 O1

Files, knowledge and retrieval

Feature	croit AIplane	LiteLLM	Open WebUI	Evidence
Upload files into a conversation	✅	⚪	✅	A18 L1 O1
Read common text and document attachments	✅	⚪	✅	A18 L1 O12
PDF text extraction	✅	⚪	✅	A18 O12
OCR for scanned documents	🟡	⚪	✅	A19 O12
Structured Office document extraction	🟡	⚪	🟡	A19 O12
Generate and edit documents in a canvas	✅	⚪	🟡	A18 O1
Restore an earlier canvas revision	✅	⚪	⚪	A18
Search across configured company knowledge	✅	⚪	✅	A10 O13
Hybrid lexical and vector retrieval	✅	⚪	✅	A10 O13
Optional cross-encoder reranking for retrieval	✅	⚪	✅	A10 O13
Search citations linked to source material	✅	⚪	✅	A10 O13
Git knowledge source	✅	⚪	⚪	A10
WebDAV knowledge source	✅	⚪	⚪	A10
Google Drive knowledge source	✅	⚪	🟡	A10 O13
HyperKitty knowledge source	✅	⚪	⚪	A10
Multiple vector database backends	🟡	⚪	✅	A10 O13
Bring-your-own embedding backend	✅	⚪	✅	A10 O13
Knowledge access by user group	✅	⚪	✅	A10 O5
Agentic retrieval through model tools	✅	⚪	✅	A10 O13
Whole-document context mode	🟡	⚪	✅	A18 O13
Manage uploaded files in a centralized file manager	⚪	⚪	✅	O14
Versioned knowledge source synchronization	✅	⚪	🟡	A10 O13
Knowledge sources limited to a configured allowlist	✅	⚪	🟡	A10 O13

Tools, integrations and extensibility

Feature	croit AIplane	LiteLLM	Open WebUI	Evidence
Built-in web search	✅	🟡	✅	A20 L15 O1
Fetch and read a web page	✅	🟡	✅	A20 L15 O1
MCP client connections	✅	✅	✅	A21 L16 O3
MCP tool gateway/proxy	🟡	✅	🟡	A21 L16 O3
Expose REST/OpenAPI tools	🟡	✅	✅	A21 L16 O3
Per-tool permission grants	✅	✅	✅	A8 L7 O5
Tool discovery for API-token callers	✅	🟡	🟡	A21 L16 O3
Tool calls from OpenAI-compatible model requests	✅	✅	✅	A21 L16 O3
Gateway executes granted server-side tools	✅	✅	🟡	A21 L16 O3
Return client-owned tool calls to the API caller	✅	✅	🟡	A21 L16 O3
Dedicated tool execution sandbox	✅	⚪	🟡	A22 O15
Per-run gVisor isolation for code execution	✅	⚪	🟡	A22 O15
Optional allowlisted sandbox network egress	✅	⚪	🟡	A22 O15
User-browser interaction through a paired extension	✅	⚪	⚪	A17
Headless sandbox browser	✅	⚪	🟡	A22 O15
Create and convert Office/PDF documents through tools	✅	⚪	🟡	A22 O15
Generate QR codes	✅	⚪	⚪	A23
Generate images through configured image backends	✅	🟡	✅	A23 L2 O1
Edit images through configured image backends	✅	🟡	✅	A23 L2 O1
Voice dictation in chat	✅	⚪	✅	A24 O1
Spoken chat / voice mode	✅	⚪	✅	A24 O1
Web Push notifications	✅	⚪	🟡	A23 O3
Connectors with managed OAuth credentials	✅	🟡	🟡	A21 L16 O3
User-managed connector accounts	✅	⚪	🟡	A21 O3
Skills as reusable Markdown instructions	✅	🟡	✅	A15 L13 O3
User-created in-process Python plugins	⚪	✅	✅	L18 O3
Request/response middleware plugin framework	⚪	✅	✅	L18 O3
Community extension catalog	⚪	🟡	✅	L18 O3

Agents and workflow automation

Feature	croit AIplane	LiteLLM	Open WebUI	Evidence
Dedicated business-agent builder	✅	🟡	🟡	A9 L19 O2
Agent-specific system instructions	✅	✅	✅	A9 L19 O2
Agent-specific model selection	✅	✅	✅	A9 L19 O2
Agent-specific knowledge grants	✅	🟡	✅	A9 L19 O2
Agent-specific tool grants	✅	✅	✅	A9 L19 O2
Agent state across runs	✅	🟡	🟡	A9 L19 O2
Agent identity and delegated permissions	✅	🟡	⚪	A9 L19
Repeatable agent test cases	✅	🟡	⚪	A25 L19
Require a passing test suite before agent publication	✅	⚪	⚪	A25
Agent version publication workflow	✅	🟡	🟡	A25 L19 O2
Publish an agent as a website widget	✅	🟡	⚪	A26 L19
Agent-to-agent (A2A) interface	✅	✅	⚪	A26 L28
Remote agent routes	✅	✅	⚪	A26 L28
Durable human approval and handoff	✅	🟡	⚪	A12 L19
Shared inbox for agent decisions	✅	⚪	⚪	A12
Collect secure user input during an agent run	✅	⚪	⚪	A12
Scheduled agent/prompt runs	✅	⚪	✅	A27 O16
Webhook-triggered agent runs	✅	🟡	🟡	A27 L20 O16
Persisted run conversation and history	✅	🟡	✅	A27 L20 O16
Agent run analytics	✅	🟡	⚪	A13 L20
Export agent activity for review	✅	⚪	⚪	A13
Agent activity hash-chain verification	✅	⚪	⚪	A13
Task lists maintained by models	⚪	⚪	✅	O1 O3
Multi-agent orchestration framework in the product	🟡	✅	🟡	A9 L19 O3

Observability, security and operations

Feature	croit AIplane	LiteLLM	Open WebUI	Evidence
Request usage records	✅	✅	✅	A8 L10 O6
Token and cost accounting	✅	✅	✅	A8 L10 O6
Per-backend health visibility	✅	✅	⚪	A5 L5
Per-backend in-flight visibility	✅	✅	⚪	A5 L5
Agent activity audit trail	✅	🟡	⚪	A13 L20
Browser-control action audit trail	✅	⚪	⚪	A17
External logging callbacks	⚪	✅	🟡	L21 O17
OpenTelemetry traces, metrics, and logs	⚪	✅	✅	L22 O18
Prometheus metrics endpoint	✅	✅	🟡	A34 L22 O18
Built-in usage dashboard	✅	✅	✅	A8 L10 O6
Blind model evaluation / arena	⚪	🟡	✅	L23 O6
Model-quality evaluation workflow	🟡	✅	✅	A25 L23 O6
Encrypted stored upstream credentials	✅	✅	🟡	A28 L24 O19
Browser session authentication	✅	✅	✅	A8 L7 O5
API credentials are not forwarded to the model provider	✅	✅	🟡	A3 L24 O19
Secret-manager integrations	🟡	🟡 Enterprise	🟡	A28 L24 L29 O19
Automated rotation of virtual API keys	⚪	🟡 Enterprise	⚪	L29
IP allowlists for gateway access	⚪	🟡 Enterprise	⚪	L29
Public/private route access controls	⚪	🟡 Enterprise	⚪	L29
Team-specific log routing and logging opt-out	⚪	🟡 Enterprise	⚪	L29
Admin-operation audit logs	⚪	🟡 Enterprise	⚪	L29
Multi-region deployment under one license	⚪	🟡 Enterprise	⚪	L25 L29
Role/group-aware authorization on application APIs	✅	✅	✅	A8 L7 O5
Public agent visitor rate limiting	✅	⚪	⚪	A29
Per-user/model/token spend budgets	✅	✅	⚪	A8 L9
Shared-state multi-replica application scaling	❌	✅	✅	A30 L25 O20
Docker deployment	✅	✅	✅	A31 L25 O20
Docker Compose deployment	✅	✅	✅	A31 L25 O20
Kubernetes Helm deployment	✅	✅	✅	A31 L25 O20
Podman/systemd Quadlet deployment	✅	⚪	⚪	A31
Terraform deployment modules	⚪	✅	⚪	L25
Source build and self-hosted runtime	✅	✅	✅	A31 L25 O20
Backup and recovery guide	✅	🟡	🟡	A32 L25 O20
Version-matched documentation served inside the installation	✅	⚪	⚪	A33
llms.txt and full Markdown manual exports	✅	⚪	⚪	A33
Signed container image verification documented	✅	✅	⚪	A35 L26
Published enterprise support/SLA offering	⚪	✅	⚪	L27

What the comparison says

AIplane's strongest position is the combination of a user-facing AI workspace, a governed application gateway, and a business-agent lifecycle in one self-hosted product. The most distinctive documented areas are browser control in a user's own browser, agent publication to a website or A2A channel, repeatable tests before publication, durable human decisions in an inbox, and version-matched documentation shipped with each installation.

Request priority gives administrators a shared Low/Medium/High setting for chat and API requests from any application. AIplane orders waiting requests by their authenticated priority, retains FIFO within each class and preserves running calls. Its queue belongs to one process; sustained higher-priority traffic can make lower classes time out. Backend forwarding depends on reported scheduling support or an explicit mapping and an enabled upstream scheduler. LiteLLM documents numeric, request-supplied priorities as a beta feature for testing only. Its reviewed scheduler guide does not establish the same group/token administration or backend translation, and the reviewed Open WebUI administration guide does not establish these scheduling capabilities; those entries remain unknown. See A36 and L31 for the operating conditions.

AIplane is behind LiteLLM on documented provider breadth, the standalone Python SDK, the batch API surface, broad third-party logging/observability integrations, Terraform deployment modules, and documented multi-replica scaling. Those are concrete gaps, not wording problems. AIplane's documented deployment topology is one application replica; do not position it as horizontally scalable today.

Open WebUI documents stronger end-user collaboration and chat customization in areas such as shared channels, multi-model side-by-side chats, a centralized file manager, and user-selectable local-password authentication. AIplane's advantage is deeper integration between employee chat, application/API governance, and business-agent lifecycle controls. The products overlap substantially in chat, knowledge retrieval, tools, voice and image workflows; check each row's conditions before calling them equivalent.

Browser control needs a narrow definition. AIplane's documented extension acts in the user's existing, authenticated browser. Open WebUI has adjacent browser and desktop capabilities, but its official Chrome-extension repository is archived and marked deprecated at the revision checked here; it points users to the separate desktop app. The reviewed current product sources do not establish that the desktop app controls the same authenticated browser tab. That is why the matrix marks the exact maintained-extension feature as absent while marking broader browser automation as partial. This is a specific AIplane advantage in the current comparison, not a claim that no Open WebUI community add-on or third-party extension can do similar work.

The marks do not establish comparative performance, security quality, or total cost. Those require deployment-specific tests, threat models and pricing evidence. “No” applies only to the named capability and documented product surface; it does not mean the vendor could not add it or that an extension could not approximate it.

AIplane's GDPR and NDA pool flags are operator declarations. They do not inspect a provider's contracts or certify compliance. LiteLLM OSS already includes virtual keys, user/team controls, budgets, spend tracking, routing, and request/response logging. Its SSO is free for up to five users; SCIM, organization/team-scoped delegated administration, team-member permission customization, model-specific budgets per key, selected built-in moderation integrations, key/team-scoped guardrails, secret managers, automated virtual-key rotation, IP allowlists, team-specific logging controls, and multi-region deployment are documented as Enterprise capabilities. This distinction matters: the matrix does not imply that LiteLLM has no access control in OSS; it separates baseline controls from paid enterprise governance. Open WebUI's feature marks describe the documented self-hosted product and can still depend on configuration or connected services.

Evidence sources

AIplane sources link to this repository's user and operator documentation. LiteLLM and Open WebUI sources link to their official public documentation. The source keys are deliberately reusable so a maintainer can update a page without repeating long URLs in every table row.

AIplane

Key	Source
A1	Deployment guide
A2	Chat guide
A3	HTTP API reference
A4	Models and routing, tool inventory
A5	Models and routing
A6	Models and routing
A7	Fallbacks and temporary outages
A8	Access, tokens and limits
A9	Agent builder, agent permissions
A10	Knowledge collections, tools and integrations, tool inventory
A11	Model settings, content guard settings
A12	Approvals and handoffs, agent decisions
A13	Agent runs and activity
A14	Account, memory and usage
A15	Tools, integrations and skills, tool inventory
A16	Language picker
A17	Tools and integrations, tool inventory
A18	Files and canvas
A19	System settings, files and documents, tool inventory
A20	Tools and integrations, tool inventory
A21	Tools and integrations, integrations administration
A22	System settings, deployment
A23	Tool inventory
A24	Files and documents, tool inventory
A25	Test and publish agents
A26	Agent channels and external agents, agent test and publication, HTTP API reference
A27	Schedules, webhooks and inbox, agent runs and inbox
A28	Operator settings, environment reference
A29	Agent visitor lifetime, rates and retained data
A30	Deployment prerequisites and scaling
A31	Deployment options
A32	Backup and recovery
A33	Versioned installed documentation
A34	Monitoring, Prometheus metrics settings
A35	Verify images, what a build publishes
A36	Request priority, group and token administration, backend forwarding

LiteLLM

Key	Official source
L1	LiteLLM README
L2	Proxy endpoints
L3	Python SDK
L4	Provider integrations
L5	Load balancing, routing and fallbacks
L6	Traffic mirroring and routing
L7	Authentication and access control
L8	Teams, users and model access
L9	Budgets and rate limits
L10	Spend tracking
L11	Guardrails
L12	Memory management
L13	Skills and agent gateway
L15	Web search and tools
L16	MCP gateway
L18	Plugins and callbacks
L19	Agent gateway
L20	Scheduled and background tasks
L21	Logging integrations
L22	Metrics and observability
L23	Evaluation
L24	Secrets and credential management
L25	Production deployment
L26	Image verification
L27	Enterprise plans and support
L28	Supported API endpoints
L29	Enterprise features and license requirements
L30	Role-based access controls, including OSS and Enterprise scopes
L31	Beta request prioritization, reviewed 11 October 2026

Open WebUI

Key	Official source
O1	Feature overview
O2	Models and agents
O3	Extensibility
O4	Model comparison and evaluation
O5	Authentication and access
O6	Administration and analytics
O7	Tools and approval controls
O8	Chat features
O9	Channels
O10	Data controls
O12	File and document handling
O13	Knowledge and RAG
O14	Central file manager
O15	Open Terminal
O16	Automations
O17	Webhooks and integrations
O18	Deployment and observability
O19	Security and data controls
O20	Deployment and scaling
O22	Official Chrome extension repository (archived and deprecated)
O23	Open WebUI Desktop
O24	Open WebUI API endpoints

Git revisions compared

The product and documentation sources were checked against these exact repository revisions. The public documentation baseline was reviewed on 4 October 2026; the request-priority update uses the 11 October 2026 documentation check described above. The revision pins identify the implementation baselines even as upstream main branches move.

Product/source	Repository	Revision
croit AIplane (implementation baseline reviewed)	croit/aiplane	e8285ffdd457ef9b5d6899e438c3484a7b8f36cf
croit AIplane (request-priority implementation reviewed)	croit/aiplane	d5b81c2abcdb77f96522c4f10b9cecf0c691c16d
LiteLLM (implementation and documentation)	BerriAI/litellm	e1d16f51d14849c1b3decf17cd81a3bcb4863dca
Open WebUI (implementation)	open-webui/open-webui	8bd8b4fac5e059578ac0c74b3c18d11139f88b7d
Open WebUI (public feature documentation)	open-webui/docs	e6e9151cae4d4aec79d1a2169475c0a41f47cad9
Open WebUI (archived official Chrome extension)	open-webui/extension	635b90c685649806b2b1e06a39c6e3e2479195d5
Open WebUI Desktop (replacement project referenced by the extension repository)	open-webui/desktop	0931b46d66e06bfb7d0b6a4aa065e6db338a1000

---

Documentation build: 2610.1.108 · commit b774e1f9e2c2