{"id":2402,"date":"2026-09-22T05:13:09","date_gmt":"2026-09-21T20:13:09","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/2026\/09\/22\/open-webui-v0-11-4\/"},"modified":"2026-09-22T05:13:09","modified_gmt":"2026-09-21T20:13:09","slug":"open-webui-v0-11-4","status":"publish","type":"post","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/22\/open-webui-v0-11-4\/","title":{"rendered":"Open WebUI v0.11.4 Released: Dramatic Docker Image Slimming"},"content":{"rendered":"<p><!-- lmw:facts --><\/p>\n<h2>At a Glance<\/h2>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Item<\/th>\n<th>Value<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Repository<\/td>\n<td><a href=\"https:\/\/github.com\/open-webui\/open-webui\">open-webui\/open-webui<\/a><\/td>\n<\/tr>\n<tr>\n<td>Version<\/td>\n<td><a href=\"https:\/\/github.com\/open-webui\/open-webui\/releases\/tag\/v0.11.4\">v0.11.4<\/a><\/td>\n<\/tr>\n<tr>\n<td>Published<\/td>\n<td>2026-09-22<\/td>\n<\/tr>\n<tr>\n<td>License<\/td>\n<td>NOASSERTION<\/td>\n<\/tr>\n<tr>\n<td>Source type<\/td>\n<td>Primary source (the publisher itself)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><em>Values determined by this site&#8217;s code at collection time. Dates are JST.<\/em><\/p>\n<p><!-- \/lmw:facts --><\/p>\n<h2>Overview<\/h2>\n<p>Open WebUI v0.11.4 has been released. Open WebUI is a Python-based project that provides a user-friendly AI interface compatible with Ollama, the OpenAI API, and more (license: NOASSERTION, GitHub stars: 152,721).<\/p>\n<p>The most impactful change for users in this update is the dramatic reduction in Docker image sizes. Notably, the &#8220;slim&#8221; image has been reduced to about 175MB, achieving a size reduction of approximately 89% compared to previous releases. This is a very valuable update for users who prioritize faster deployments and storage savings.<\/p>\n<h2>Breaking Changes &amp; Deprecations<\/h2>\n<p>Along with the Docker image slimming, some tools and packages are no longer pre-installed. Depending on your current setup, please note that functionality might break after updating.<\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th style=\"text-align: left;\">Change<\/th>\n<th style=\"text-align: left;\">Old<\/th>\n<th style=\"text-align: left;\">New<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td style=\"text-align: left;\">Slim image composition<\/td>\n<td style=\"text-align: left;\">Included local models and toolsets<\/td>\n<td style=\"text-align: left;\">Removed (slimmed down to approx. 175MB)<\/td>\n<\/tr>\n<tr>\n<td style=\"text-align: left;\">Pre-installed packages<\/td>\n<td style=\"text-align: left;\">Included <code>nltk<\/code>, <code>google-api-python-client<\/code>, etc.<\/td>\n<td style=\"text-align: left;\">Removed (excluded from dependencies)<\/td>\n<\/tr>\n<tr>\n<td style=\"text-align: left;\"><code>uv<\/code> command inside Docker<\/td>\n<td style=\"text-align: left;\">Available inside the container<\/td>\n<td style=\"text-align: left;\">Removed (used only during build)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<ul>\n<li><strong>Slim image users<\/strong>: Local models, packages, and tools have been removed. If you depended on these, you need to migrate to the standard image or install them manually.<\/li>\n<li><strong>Custom tool\/function developers<\/strong>: Packages related to <code>nltk<\/code> and the Google API have been removed. Tools and functions using these must explicitly list them in the front matter <code>requirements<\/code> to have them installed at runtime.<\/li>\n<\/ul>\n<h2>Key Changes<\/h2>\n<h3>Dramatic Docker Image Slimming<\/h3>\n<p>The slim image is now approximately 175MB (an ~89% reduction) by removing local models and related tools. The standard image has also been reduced by about 170MB by cleaning up redundant Python environments, unused fonts, and packages. This makes pulling and starting containers much faster.<\/p>\n<h3>Terminal Integration and Skill Enhancements<\/h3>\n<p>&#8220;Skills&#8221; provided by connected terminal servers can now be used directly from menus (&#8220;$&#8221; or &#8220;\/&#8221;), similar to workspace skills. Additionally, a feature has been added where models automatically read <code>AGENTS.md<\/code> in the terminal&#8217;s home directory, saving the effort of manually pasting instructions. Furthermore, running commands in the terminal now have separate tabs, making it easier to monitor multiple processes in parallel.<\/p>\n<h3>Performance Improvements for Long Chats and Redis Usage<\/h3>\n<p>Processing during long conversation streaming, where the entire conversation was re-read and written for each message, has been optimized. This prevents processing loads from escalating even as conversations grow longer. In addition, model lists are now cached by each worker in Redis shared environments, significantly reducing synchronization costs in environments with large numbers of models.<\/p>\n<h3>Non-Blocking Search and File Operations Improvements<\/h3>\n<p>Searches (grep) within chats and knowledge bases are now executed outside the event loop (in a separate thread). This ensures the server keeps running smoothly without making other requests wait, even while executing large-scale searches. In addition, the terminal file browser now allows selecting two files to compare side-by-side (diff view), and supports bulk folder uploads.<\/p>\n<h3>Expanded UI Customization and Multilingual Support<\/h3>\n<p>Model and tool names, descriptions, and starter prompts can now be set per language. Administrators can also directly rewrite interface text by language from the settings panel. Furthermore, small usability improvements have been added, such as the ability to set background images for each workspace model and a shortcut to execute deletion with a single click while holding down the Shift key.<\/p>\n<h3>Enhanced Security and Access Control<\/h3>\n<p>A feature has been added to immediately terminate all sessions upon sign-out. Restrictions have also been introduced to prohibit SVG and Mermaid diagrams from referencing external origins, reducing security risks. Role checks have been introduced to model list fetching endpoints for specific connections (Ollama\/OpenAI), tightening permission management.<\/p>\n<h2>Strengths and Use Cases<\/h2>\n<p>Open WebUI is not an inference engine relying on a specific model architecture; rather, it is an interface to utilize a wide variety of large language models (LLMs) via Ollama or OpenAI API-compatible servers connected as a backend. In this v0.11.4 release, the following features have been added and organized for model operations and integration with external services.<\/p>\n<p>For hardware acceleration, official support is provided for CUDA environments to utilize Nvidia GPU. A <code>:cuda<\/code> tag is available as a Docker image, enabling fast GPU-based inference in Linux or WSL environments with the Nvidia CUDA container toolkit installed. For environments without a GPU, standard CPU-only images and images tagged with <code>:ollama<\/code> (which bundle Ollama inside the container) are also provided.<\/p>\n<p>For external integration features, &#8220;Staan&#8221; is now selectable as a web search provider for users requiring searches within the European Union (EU). This can be configured through the web search settings in the admin panel or via the <code>STAAN_API_KEY<\/code> environment variable. Additionally, for existing Exa searches, the maximum character length per result can now be limited using <code>EXA_MAX_CONTENT_LENGTH<\/code>, helping to curb context window consumption.<\/p>\n<p>Regarding RAG (Retrieval-Augmented Generation) features, functionality has been added to retain custom metadata attached to uploaded files along with the split chunks, passing them to the model as search results. Administrators can use <code>RAG_SOURCE_METADATA_KEYS<\/code> to control which metadata fields are presented to the model.<\/p>\n<p>In terms of model management, improvements have been made so that if a specific model is removed from the connection target or temporarily goes offline, chats will not freeze with &#8220;Model Not Selected&#8221;. A mechanism has been introduced to automatically fallback to the user&#8217;s default model, the admin-configured default, or the first available model, improving convenience when continuing past conversations. Furthermore, a filtering option has been added to the admin panel model list to distinguish between base models and models created in the workspace.<\/p>\n<h2>How to Get It<\/h2>\n<p>Update Open WebUI according to your installation method using the steps below.<\/p>\n<h3>Python Environment (pip)<\/h3>\n<p>If you installed using Python&#8217;s pip, run the following commands. <strong>Python 3.11<\/strong> is recommended to avoid compatibility issues.<\/p>\n<pre><code class=\"language-bash\">pip install open-webui\nopen-webui serve\n<\/code><\/pre>\n<h3>Docker Environment<\/h3>\n<p>Pull the latest image and restart the container. Make sure the <code>-v open-webui:\/app\/backend\/data<\/code> option is included for data persistence.<\/p>\n<p><strong>Standard startup command when Ollama is running on the same host:<\/strong><\/p>\n<pre><code class=\"language-bash\">docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:\/app\/backend\/data --name open-webui --restart always ghcr.io\/open-webui\/open-webui:main\n<\/code><\/pre>\n<p><strong>To enable Nvidia GPU (CUDA) support:<\/strong><\/p>\n<pre><code class=\"language-bash\">docker run -d -p 3000:8080 --gpus all --add-host=host.docker.internal:host-gateway -v open-webui:\/app\/backend\/data --name open-webui --restart always ghcr.io\/open-webui\/open-webui:cuda\n<\/code><\/pre>\n<p><strong>To use an image with bundled Ollama (using GPU):<\/strong><\/p>\n<pre><code class=\"language-bash\">docker run -d -p 3000:8080 --gpus=all -v ollama:\/root\/.ollama -v open-webui:\/app\/backend\/data --name open-webui --restart always ghcr.io\/open-webui\/open-webui:ollama\n<\/code><\/pre>\n<p><!-- lmw:releases --><\/p>\n<h2>Releases Since Our Last Article<\/h2>\n<p><em>Compiled by Local Model Watch from the project&#8217;s GitHub releases: the versions between this release and the last one we covered, which did not get separate articles.<\/em> <em>Full history: <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-open-webui-en\/\">release tracker<\/a>.<\/em><\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Version<\/th>\n<th>Released<\/th>\n<th>Release notes<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>v0.11.3<\/td>\n<td>2026-08-31<\/td>\n<td><a href=\"https:\/\/github.com\/open-webui\/open-webui\/releases\/tag\/v0.11.3\">GitHub<\/a><\/td>\n<\/tr>\n<tr>\n<td>v0.11.2<\/td>\n<td>2026-08-31<\/td>\n<td><a href=\"https:\/\/github.com\/open-webui\/open-webui\/releases\/tag\/v0.11.2\">GitHub<\/a><\/td>\n<\/tr>\n<tr>\n<td>v0.11.1<\/td>\n<td>2026-08-26<\/td>\n<td><a href=\"https:\/\/github.com\/open-webui\/open-webui\/releases\/tag\/v0.11.1\">GitHub<\/a><\/td>\n<\/tr>\n<tr>\n<td>v0.11.0<\/td>\n<td>2026-07-27<\/td>\n<td><a href=\"https:\/\/github.com\/open-webui\/open-webui\/releases\/tag\/v0.11.0\">GitHub<\/a><\/td>\n<\/tr>\n<tr>\n<td>v0.10.2<\/td>\n<td>2026-07-01<\/td>\n<td><a href=\"https:\/\/github.com\/open-webui\/open-webui\/releases\/tag\/v0.10.2\">GitHub<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><!-- \/lmw:releases --><\/p>\n<p><!-- lmw:related --><\/p>\n<h2>Related Articles<\/h2>\n<ul>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/12\/litelm-lightweight-llm-routing-library\/\">litelm: Lightweight LLM Routing and Message Conversion Library<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:related --><\/p>\n<h2>Sources<\/h2>\n<ul>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/releases\/tag\/v0.11.4\">https:\/\/github.com\/open-webui\/open-webui\/releases\/tag\/v0.11.4<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/29731\">https:\/\/github.com\/open-webui\/open-webui\/pull\/29731<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/29723\">https:\/\/github.com\/open-webui\/open-webui\/pull\/29723<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/29725\">https:\/\/github.com\/open-webui\/open-webui\/pull\/29725<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/29726\">https:\/\/github.com\/open-webui\/open-webui\/pull\/29726<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/29728\">https:\/\/github.com\/open-webui\/open-webui\/pull\/29728<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/29757\">https:\/\/github.com\/open-webui\/open-webui\/pull\/29757<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/29621\">https:\/\/github.com\/open-webui\/open-webui\/pull\/29621<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/28185\">https:\/\/github.com\/open-webui\/open-webui\/pull\/28185<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/30138\">https:\/\/github.com\/open-webui\/open-webui\/pull\/30138<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/open-webui\/open-webui\/pull\/29499\">https:\/\/github.com\/open-webui\/open-webui\/pull\/29499<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Open WebUI v0.11.4 is released with a dramatic reduction in Docker image size, terminal skill integration, and performance improvements.<\/p>\n","protected":false},"author":1,"featured_media":2401,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[105],"tags":[1475,1093,1917,590,1547],"class_list":["post-2402","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-engines-and-tools","tag-docker-en","tag-ollama-en","tag-open-webui-en","tag-python-en","tag-verified"],"lang":"en","translations":{"en":2402,"ja":2400},"_links":{"self":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/2402","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/comments?post=2402"}],"version-history":[{"count":0,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/2402\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media\/2401"}],"wp:attachment":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media?parent=2402"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/categories?post=2402"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/tags?post=2402"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}