{"id":20274,"date":"2026-11-19T08:09:26","date_gmt":"2026-11-18T23:09:26","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/2026\/11\/19\/comfyui-v0460-released\/"},"modified":"2026-11-19T08:09:26","modified_gmt":"2026-11-18T23:09:26","slug":"comfyui-v0460-released","status":"publish","type":"post","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/19\/comfyui-v0460-released\/","title":{"rendered":"ComfyUI v0.46.0 Released: DirectML Support Dropped"},"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\/Comfy-Org\/ComfyUI\">Comfy-Org\/ComfyUI<\/a><\/td>\n<\/tr>\n<tr>\n<td>Version<\/td>\n<td><a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/releases\/tag\/v0.46.0\">v0.46.0<\/a><\/td>\n<\/tr>\n<tr>\n<td>Published<\/td>\n<td>2026-11-19<\/td>\n<\/tr>\n<tr>\n<td>License<\/td>\n<td>GPL-3.0<\/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 when the information was collected. Dates are JST.<\/em><\/p>\n<p><!-- \/lmw:facts --><\/p>\n<h2>Overview<\/h2>\n<p>ComfyUI v0.46.0 has been released. ComfyUI is a GUI, API, and backend for diffusion models featuring a graph\/node-based interface, written in Python and licensed under GPL-3.0.<\/p>\n<p>The most impactful change in this version is the <strong>removal of DirectML support<\/strong> (PR #17711). Users running ComfyUI via DirectML should check for alternatives before updating. In other environments, the update focuses on added support and bug fixes surrounding Qwen and SenseNova.<\/p>\n<h2>Breaking Changes and Deprecations<\/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>Old<\/th>\n<th>New<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>DirectML support available<\/td>\n<td>DirectML support removed (PR #17711)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>The release notes contain only a single line stating &#8220;Remove directml support,&#8221; and no migration paths or alternative procedures are documented. Users running ComfyUI with DirectML may find that it stops working after the update. In that case, it is safer to stick with v0.45.0.<\/p>\n<h2>Key Changes<\/h2>\n<h3>Warning When CUDA 13 Is Not Used<\/h3>\n<p>If an NVIDIA GPU supports CUDA 13 but it is not being used, a warning is now displayed (PR #17760, changed to yellow in #17769). The README states that PyTorch with cu130 or higher is required for NVIDIA 20 series and newer. Users who see this warning should consider updating PyTorch.<\/p>\n<h3><code>--force-output-rerun<\/code> Option<\/h3>\n<p>This option forces output nodes to write files again even when there is a cache hit (PR #17809). It is intended for users who were troubled by files not being saved when re-running with the same input. This does not change the default behavior and is only applied as an option when needed.<\/p>\n<h3>Text Generation and Text Encoders<\/h3>\n<ul>\n<li><strong>SenseNova<\/strong>: Added support for thinking, interleaved generation, and self-correction (PR #16032). Text generation has also been accelerated (PR #17811).<\/li>\n<li><strong>Qwen3-1.7B<\/strong>: Now available for use in the Generate Text node (PR #17787).<\/li>\n<li><strong>Qwen Vocabulary-Reduced Checkpoints<\/strong>: Added support for embedded tokenizers (PR #17633).<\/li>\n<li><strong>PAN2<\/strong>: Added support (PR #17755).<\/li>\n<\/ul>\n<p>None of these include detailed documentation, so users of the target models will need to test them after updating. There is no information regarding whether re-conversion or re-downloading is required.<\/p>\n<h3>Video and Image Nodes<\/h3>\n<ul>\n<li>Added a Video Deflicker node and an accelerated RAFT optical flow implementation (PR #17754).<\/li>\n<li>Added a new node to split image channels (PR #13548).<\/li>\n<li>The Save Image Advanced node can now save single-channel PNGs (PR #17727).<\/li>\n<li>Added a mesh input to the Create Camera Info node, allowing arbitrary meshes to be framed (PR #17730).<\/li>\n<li>Concatenate Video clips can now match the frame rate and audio of the first clip (PR #17621). This may change the concatenation results of existing workflows.<\/li>\n<\/ul>\n<h3>Bug Fixes<\/h3>\n<ul>\n<li>Fixed an issue where some workflows would not run properly (PR #17675).<\/li>\n<li>Fixed an issue where apg could affect outputs when cfg was 1.0 (PR #17741).<\/li>\n<li>Fixed an issue during dynamic VRAM buffer restoration where sampling schedules leaked between models (PR #16491). This is relevant for users switching between multiple models.<\/li>\n<\/ul>\n<h3>Partner Nodes<\/h3>\n<p>API nodes for LTX, MiniMax, Kling 4.0, and Seedance 2.5 Flash were added, while Wan, Krea, and Luma migrated to Comfy Router. These nodes use external API services and do not directly affect local inference.<\/p>\n<h2>Supported Models and Hardware<\/h2>\n<p>Support for new GPUs or accelerators is not mentioned in the documentation. Regarding models, the following support has been added:<\/p>\n<ul>\n<li><strong>SenseNova<\/strong>: Generation using thinking, interleaved generation, and self-correction is now supported (PR #16032), along with accelerated text generation (PR #17811).<\/li>\n<li><strong>Qwen3-1.7B<\/strong>: Text generation using Qwen3-1.7B is now available in the Generate Text node (PR #17787).<\/li>\n<li><strong>Vocabulary-Reduced Qwen Checkpoints<\/strong>: Added support for cases where the tokenizer is embedded in the checkpoint (PR #17633).<\/li>\n<li><strong>PAN2<\/strong>: Added support (PR #17755).<\/li>\n<\/ul>\n<p>In all cases, the PR bodies are not included in the documentation. Because the acquisition sources for corresponding model files, required placement directories, and conversion requirements cannot be verified from the documentation, they are not listed here. When using them, please check workflow templates or official documentation after updating the main ComfyUI installation.<\/p>\n<h2>How to Get It<\/h2>\n<p>The README excerpt in the documentation does not contain specific commands to update the ComfyUI core to v0.46.0. Therefore, please check the <a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/releases\/tag\/v0.46.0\">release page<\/a> for instructions. The differences can be viewed in the <a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/compare\/v0.45.0...v0.46.0\">v0.45.0&#8230;v0.46.0 Full Changelog<\/a>.<\/p>\n<p>For peace of mind, it is recommended to review the hardware requirements listed in the README before updating.<\/p>\n<ul>\n<li>Python 3.13 is fully supported. Python 3.14 also works, but may cause issues with some custom nodes.<\/li>\n<li>PyTorch 2.7 has minimal support, and using newer versions is strongly recommended.<\/li>\n<li>PyTorch with cu130 or higher is required for NVIDIA 20 series and newer.<\/li>\n<\/ul>\n<p>Note that <code>pip install -r manager_requirements.txt<\/code> and <code>python main.py --enable-manager<\/code> found in the README are steps to enable the Manager, not instructions for updating the core application.<\/p>\n<p><!-- lmw:related --><\/p>\n<h2>Related Articles<\/h2>\n<ul>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/comfyui-v0450-released\/\">ComfyUI v0.45.0 Released with Qwen 3.x Optimizations<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/comfyui-v0440-released\/\">ComfyUI v0.44.0 Released with Expanded Partner Nodes<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/comfyui-v0-43-0-released\/\">ComfyUI v0.43.0 Released: MOSS-Image and Fibo 2.0 Support<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/comfyui-v0420-released\/\">ComfyUI v0.42.0 Released: MOSS-image Support and Qwen Optimizations<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:related --><\/p>\n<p><!-- lmw:next-steps --><\/p>\n<h2>What to Read Next<\/h2>\n<ul>\n<li><strong>Follow this tool<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-comfyui-en\/\">ComfyUI overview and release history (164 releases tracked)<\/a><\/li>\n<li><strong>Other image, video and audio<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-diffusers-en\/\">Diffusers<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-invokeai-en\/\">InvokeAI<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-whisper-cpp-en\/\">whisper.cpp<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:next-steps --><\/p>\n<h2>Sources<\/h2>\n<ul>\n<li><a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/releases\/tag\/v0.46.0\">Comfy-Org\/ComfyUI v0.46.0 Release Notes<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/compare\/v0.45.0...v0.46.0\">Full Changelog: v0.45.0&#8230;v0.46.0<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/pull\/16032\">PR #16032 SenseNova Support<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/pull\/17811\">PR #17811 SenseNova Text Generation Acceleration<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/pull\/17787\">PR #17787 Qwen3-1.7B Support<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/pull\/17633\">PR #17633 Qwen Embedded Tokenizer Support<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/Comfy-Org\/ComfyUI\/pull\/17755\">PR #17755 PAN2 Support<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>ComfyUI v0.46.0 is out, featuring the removal of DirectML support, new Qwen and SenseNova integrations, and various bug fixes.<\/p>\n","protected":false},"author":1,"featured_media":20273,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[105],"tags":[677,5416,5418,375,5344,1547],"class_list":["post-20274","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-engines-and-tools","tag-comfyui-en","tag-directml-en","tag-gpl-3-0-en","tag-qwen3-en","tag-sensenova-en","tag-verified"],"lang":"en","translations":{"en":20274,"ja":20272},"_links":{"self":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/20274","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=20274"}],"version-history":[{"count":0,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/20274\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media\/20273"}],"wp:attachment":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media?parent=20274"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/categories?post=20274"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/tags?post=20274"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}