InvokeAI v6.14.3 Released: Editable Vector Layers & More

At a Glance
| Item | Value |
|---|---|
| Repository | invoke-ai/InvokeAI |
| Version | v6.14.3 |
| Published | 2026-11-02 |
| License | Apache-2.0 |
| Source type | Primary source (the publisher itself) |
Values determined by this site’s code when the information was collected. Dates are JST.
Overview
The latest patch version of the image generation tool “InvokeAI", “v6.14.3", has been released. InvokeAI is a creative engine equipped with a WebUI that allows users to intuitively operate image generation models such as Stable Diffusion in their local environment.
The biggest change in this update is the introduction of editable vector layers to Canvas, the integrated canvas feature. Vector shapes and splines can now be freely adjusted on the canvas, rasterized, and passed to img2img, improving the controllability of compositions and shapes in local image generation workflows. It also includes bug fixes for resuming downloads in the model management feature, making this a highly convenient update for creators utilizing Canvas and users who frequently acquire additional models.
Key Changes
Addition of Editable Vector Layers and Splines to Canvas
A feature has been added to the Canvas function to create vector graphics layers where editable shapes and splines can be placed. This contribution is by contributor DustyShoe.
With this feature, users can create new vector layers on Canvas and place and adjust editable shapes and splines. Once the placed vector shapes achieve the intended finish, rasterization (bitmap conversion) can be executed, making it possible to incorporate them directly into the img2img generation workflow as input images.
This change affects all users who sketch out compositions on Canvas or create generation bases for img2img using smooth lines and shapes. It eliminates the trouble of creating and importing rough sketches using external drawing software in advance, and allows vector shapes to be edited and rasterized directly inside the WebUI, facilitating iterative trial and error in image generation.
No tasks such as re-downloading models or re-converting existing data are required for use. It can be used by launching the WebUI after the update and creating a new vector layer from the Canvas layer features.
Bug Fix Allowing Interrupted Model Downloads to Resume
A bug has been fixed in Model Manager where partially downloaded files could not be resumed after a server crash or termination occurred. This contribution is by contributor primorLee.
Before the fix, if the server process terminated abnormally or was stopped manually while downloading a large model, a phenomenon occurred where the restored download task could not be resumed properly. With this fix, even after unexpected interruptions occur, downloads that were acquired halfway through can now be kept and restarted.
This change affects all users downloading new model files to their local environment through Model Manager. Especially when downloads take time due to network connection reasons, or in the event of a PC restart or unexpected crash, there is no need to restart the download from the beginning.
Users do not need to perform any special settings changes or advance preparations. Simply updating to this version allows the download restoration and resumption process upon server restart to function properly.
Backport of Weblate Translations for v6
Multilingual UI translation resources updated on Weblate have been backported for the v6 lineage. It has been incorporated by lstein as PR #9878.
This change affects users utilizing the WebUI in language environments other than English. UI display text translations are reflected up to date. No additional user-side settings are required for use; simply applying the update reflects the latest UI translations.
How to Get It
To install and update InvokeAI, use the officially provided launcher (Launcher).
In the official installation guide, procedures for obtaining the latest launcher from the link below are provided. When updating an existing environment or building a new environment, obtain the latest version from here and proceed with setup.
Distribution status of binaries and installers for each platform, as well as detailed release information, can be checked directly from the GitHub releases page.
Related Articles
- InvokeAI v6.14.2 Released with Custom Fonts and Model Fixes
- InvokeAI v6.14.1 Released with Krea-2 LoRA and Memory Fixes
What to Read Next
- Follow this tool → InvokeAI overview and release history (116 releases tracked)
- Other image, video and audio → ComfyUI / Diffusers / whisper.cpp
Sources
- GitHub Releases: invoke-ai/InvokeAI v6.14.3
- GitHub PR #9194: feat(canvas): add editable vector layers and spline workflows
- GitHub PR #9521: fix(model-install): keep restored downloads restartable
- GitHub PR #9878: chore(ui): back-port Weblate translations to v6
- GitHub PR #9705: ci(docs): deploy invoke.ai from v6-bugfixes
- GitHub PR #9640: chore(prep): bump version
- GitHub Compare: v6.14.2…v6.14.3
- GitHub Releases: invoke-ai/launcher

