From 5512370f0270073193a5285f42dec0eb2e345d49 Mon Sep 17 00:00:00 2001 From: Classic298 <27028174+Classic298@users.noreply.github.com> Date: Thu, 29 Oct 2026 09:44:34 +0100 Subject: [PATCH] docs: add faster chat search to 0.12.5 changelog --- CHANGELOG.md | 1 + 1 file changed, 1 insertion(+) diff --git a/CHANGELOG.md b/CHANGELOG.md index 273d604bfd..715d10b7be 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -23,6 +23,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0 - โฑ๏ธ **Terminal start timeout.** Open WebUI now waits up to 60 seconds by default when it asks a terminal for its skills, AGENTS.md file or uploads, so a terminal that is still starting can answer, with "AIOHTTP_CLIENT_TIMEOUT_TERMINAL" to change the limit and a message naming the terminal when it still times out. [Commit](https://github.com/open-webui/open-webui/commit/ab133dba2183491d7c22229ae2b3b30cacf020d4) - ๐Ÿšจ **Automation failure notifications.** Automation run failed is now an event users can pick for their notification targets under Settings > Notifications, so the owner hears about every failed run with the automation's name and error. [#33331](https://github.com/open-webui/open-webui/pull/33331), [#33228](https://github.com/open-webui/open-webui/issues/33228), [Commit](https://github.com/open-webui/open-webui/commit/e49ff5df64a1523b80fce8ef03637953265c4355) - โšก **Faster knowledge base commands.** Knowledge base commands such as ls, cat and grep and the knowledge base sync check now list files without loading every file's full text, so on large knowledge bases each call is up to ten times faster and uses far less memory. [#33332](https://github.com/open-webui/open-webui/pull/33332), [#33322](https://github.com/open-webui/open-webui/issues/33322), [Commit](https://github.com/open-webui/open-webui/commit/517f824eb42615b203eb5c1f8779c268aa829960) +- ๐Ÿ”ญ **Faster chat search.** Searching your chats, including from the Reference Chats menu and the model's chat search tool, no longer loads every matching chat in full, so results come back faster on large chat histories, and a result's message preview in the Reference Chats menu now loads when you hover it. [Commit](https://github.com/open-webui/open-webui/commit/44a9d499db851f7c6243ada8bfddb9d5b419f473) - ๐Ÿ“ฆ **Batched document embedding.** Large documents are now embedded and saved in batches of 1000 chunks by default, set with "RAG_EMBEDDING_BUFFER_SIZE", so processing a very large file holds far less in memory at once. [Commit](https://github.com/open-webui/open-webui/commit/58536b65038dc7ff17d02dd78c60cc71ce110f92) - ๐Ÿงพ **Mistral OCR model choice.** Administrators can now pick the Mistral OCR model and send extra request options through a Parameters field in Admin Settings > Documents or "MISTRAL_OCR_PARAMS", for example to stay on an older OCR model or to match what an Azure AI Foundry endpoint expects. [#33333](https://github.com/open-webui/open-webui/pull/33333), [#19707](https://github.com/open-webui/open-webui/issues/19707), [Commit](https://github.com/open-webui/open-webui/commit/f1f0a194aef3d029f8243f7e7ac553d492283c07) - ๐Ÿ“‰ **Negative relevance threshold.** The Relevance Threshold in Admin Settings > Documents now takes negative values for rerankers that return raw scores below zero, where the field refused them and a negative "RAG_RELEVANCE_THRESHOLD" kept the page from saving while hybrid search was on. [#33330](https://github.com/open-webui/open-webui/pull/33330), [#29888](https://github.com/open-webui/open-webui/issues/29888), [Commit](https://github.com/open-webui/open-webui/commit/34ef99dbfb0815fbca10d3faea2da7c50f29e19f)