Open WebUI

Provides a single web UI that routes requests to multiple LLM backends (OpenAI, Anthropic, Ollama, LM Studio, etc.) through a pluggable provider abstraction layer. Implements model registry pattern with dynamic provider detection, allowing users to swap or add backends without code changes. Supports streaming responses, token counting, and cost tracking across heterogeneous model families.

Delivers a full-featured web UI (React/TypeScript frontend) that runs entirely on user infrastructure without external dependencies or cloud callbacks. Uses service workers and local storage for offline capability, caching conversation history and model metadata locally. Frontend communicates with backend via REST/WebSocket APIs, enabling deployment on any Docker-compatible environment or bare metal.

Integrates web search capabilities (via SearXNG, Google Search API, or Brave Search) to augment LLM responses with current information. Implements automatic search triggering based on query analysis (detects questions requiring real-time data) or manual user-initiated search. Search results are ranked by relevance and automatically injected into LLM context as augmented prompts. Supports search result caching to avoid redundant queries.

Integrates image generation models (Stable Diffusion, DALL-E, Midjourney) and vision models (GPT-4V, Claude Vision, LLaVA) into the chat interface. Supports image generation from text prompts with model-specific parameters (guidance scale, steps, sampler). Vision models can analyze uploaded images and answer questions about them. Generated images are stored locally and can be referenced in subsequent prompts.

Provides a library of reusable prompt templates with variable placeholders and conditional logic. Templates support Jinja2-style variable substitution, allowing dynamic prompt generation based on user input or conversation context. Includes built-in templates for common tasks (summarization, translation, code review) and supports custom template creation. Templates can be organized into categories and shared across users.

Enables side-by-side comparison of responses from multiple models on the same prompt. Implements A/B testing infrastructure to systematically compare model outputs with user ratings and feedback. Stores comparison results for analysis and model selection optimization. Supports blind testing (user doesn't know which model generated which response) to reduce bias. Generates comparison reports with metrics (response quality, speed, cost).

Integrates vector embedding and semantic search capabilities to enable retrieval-augmented generation (RAG) workflows. Supports document upload (PDF, TXT, Markdown), automatic chunking with configurable overlap, and embedding generation via local or remote embedding models. Uses vector database abstraction (supports Chroma, Weaviate, Milvus) to store and retrieve semantically similar chunks, injecting relevant context into LLM prompts automatically.

Maintains multi-turn conversation history with automatic context windowing and optional summarization. Stores conversations in local database (SQLite by default) with full-text search indexing. Implements sliding context window to manage token limits — automatically truncates or summarizes older messages when approaching model token limits. Supports conversation branching and editing of past messages to explore alternative response paths.

Provides a schema-based function registry that maps LLM function-calling outputs to executable Python functions or HTTP endpoints. Supports OpenAI-style function calling, Anthropic tool_use, and custom tool definitions via JSON schema. Implements automatic parameter validation, error handling with retry logic, and response injection back into conversation context. Tools can be chained sequentially or in parallel depending on LLM capabilities.

Supports fine-tuning of local models using conversation data or custom datasets. Implements dataset preparation (conversation export to training format), parameter-efficient fine-tuning via LoRA (Low-Rank Adaptation), and model merging. Fine-tuned models are saved locally and can be registered as new model variants in the provider registry. Includes evaluation metrics (perplexity, BLEU) for fine-tuning quality assessment.

Provides a plugin architecture allowing developers to extend Open WebUI with custom actions, filters, and middleware. Plugins are JavaScript/TypeScript modules that hook into request/response pipelines, conversation events, and UI rendering. Supports plugin marketplace for community-contributed extensions. Each plugin declares dependencies, permissions, and configuration schema, enabling sandboxed execution and permission-based access control.

Supports multi-user deployments with user accounts, authentication, and role-based access control (RBAC). Implements user roles (admin, user, guest) with granular permissions (model access, conversation sharing, plugin installation). Uses JWT tokens for stateless authentication and session management. Supports LDAP/Active Directory integration for enterprise deployments. Each user has isolated conversation history and model preferences.

Provides UI controls for adjusting LLM inference parameters (temperature, top_p, top_k, max_tokens, repetition_penalty) with real-time preview of parameter effects. Implements model-specific parameter presets (e.g., 'creative', 'precise', 'balanced') that optimize for different use cases. Supports quantization profiles for local models to trade off quality for speed/memory. Includes token counting and cost estimation for cloud models.

Supports exporting conversations in multiple formats (JSON, Markdown, PDF, HTML) with configurable content inclusion (messages, metadata, timestamps, model info). Implements format converters that transform conversation structure into target format, preserving formatting and metadata. Supports batch export of multiple conversations. Exported conversations can be re-imported to restore state or shared with external systems.