O narzędziu
Open WebUI — self-hosted AI Platform. 136K+ GitHub stars (z 128K w III 2026). User-friendly AI Interface (Ollama, OpenAI API, dziesiątki providers). Pipelines framework — UI-agnostic OpenAI-compatible plugin (modular workflows). Recent: Mistral TTS, STT preprocessing bypass, knowledge tool hybrid search, delete Ollama models z UI, backend outlet filters. RAG, MCP support, multi-user.
Zastosowanie
- •Self-hosted ChatGPT-like UI dla teams.
- •RAG aplikacje z hybrid search.
- •Multi-user company-wide LLM platform.
- •Custom workflows przez Pipelines.
- •Local + cloud LLM agregator (Ollama + OpenAI).
Funkcje 2026
- •Mistral TTS support (text-to-speech provider).
- •STT preprocessing bypass (AUDIO_STT_SKIP_PREPROCESSING).
- •Knowledge tool hybrid search (query_knowledge_files z hybrid + reranking).
- •Delete Ollama models bezpośrednio z UI.
- •Backend outlet filters (reliable w backend completion flows).
Funkcje główne
- •Multi-provider support: Ollama, OpenAI API, Anthropic, Gemini, Azure i więcej.
- •Multi-user features.
- •RAG (chat z documents).
- •MCP tool support.
- •AI Plans dla multi-step workflows.
- •Knowledge bases.
- •User permissions i RBAC.
- •Theming i customization.
Funkcje dodatkowe
▶Mistral TTS Support (NEW 2026)
Wsparcie dla Mistral TTS (text-to-speech) provider — voice synthesis dla chat responses. Plus STT preprocessing bypass (AUDIO_STT_SKIP_PREPROCESSING) dla raw audio uploads.
▶Pipelines Framework
UI-agnostic OpenAI-compatible plugin framework — modular, customizable workflows. Few lines of code by dodac custom logic, filters, integrations. Separate service dla complex orchestration.
▶Multi-provider Support
Ollama, OpenAI API, Anthropic, Gemini, Azure i wiecej w jednym interfejsie. Pozwala teamom uzywac mieszanki local i cloud modeli bez przelaczania interfaces.
▶Multi-user + RBAC
Multi-user platform z Role-Based Access Control. Pozwala na company-wide LLM platform z separated user permissions, audit logs i admin controls.
▶RAG z Hybrid Search + Reranking
Knowledge tool z hybrid search (query_knowledge_files) i reranking. Najlepszy w branzy open-source UI RAG quality — konkurencyjny z proprietary enterprise solutions.
▶MCP Tool Support
Native support dla Model Context Protocol tools — eksponowanie external tools (filesystem, APIs, databases) dla agents. Backend outlet filters reliable w completion flows.
▶AI Plans
Multi-step workflows z AI Plans — chain prompts, conditional logic, tool calls. Pozwala na automation dla typical use cases bez koniecznosci pisania kodu.
▶Knowledge Bases
Wbudowane knowledge bases dla RAG i context — upload documents, organize w collections, share w teamie. Single source of truth dla company-wide AI knowledge.
▶Functions/Tools/Filters
Extensions framework: Functions (custom processing), Tools (external integrations), Filters (modify input/output). Modularny system rozszerzen bez modyfikacji core.
▶Native Ollama + Docker/K8s
Native Ollama integration — local LLM zero config. Docker i Kubernetes deployment dla enterprise scale. 136K+ GitHub stars (z 128K w III 2026) — najbardziej popularna open-source UI.
✓ Zalety
Cennik
- •Open-source (BSD).
- •$0.
- •GitHub: open-webui/open-webui.
- •136K+ stars (z 128K w III 2026).
API i integracje
- •OpenAI API compatible.
- •Pipelines (Python plugin framework).
- •Functions, Tools, Filters extensions.
- •Native Ollama integration.
- •Docker, Kubernetes deployment.
Pipelines framework
- •UI-agnostic OpenAI-compatible plugin framework.
- •Modular customizable workflows.
- •Easy extend functionalities, integrate logic, dynamic workflows.
- •Few lines of code.
- •Separate service (complex orchestration).