octos Releases

This release focuses on dependency management and packaging improvements for the project.

Feb 12, 2026

This release introduces user interface improvements for data persistence and adds configuration capabilities for Model Context Protocol (MCP) servers.

Feb 6, 2026

Added support for German and French lip-sync language modules and introduced a new `isRaw` option in `speakAudio` to skip internal breaks, alongside minor fixes.

Dec 8, 2025

Added an Alt-Tab recent windows switcher, implemented true Wayland maximize, improved fullscreen and maximize animations, and added horizontal window dragging to scroll.

Nov 29, 2025

Improved streaming stability and AudioWorklet reuse, added an experimental avatar-only mode, and exposed internal properties for animation control.

Sep 3, 2025

Introduces a new logo, a new wiki interface with interactive search, xwayland-satellite integration for X11 apps, and basic screen reader support for accessibility.

Aug 30, 2025

Initial release of Octos, an HTML-based dynamic desktop engine for Windows that enables interactive wallpapers with multi-monitor support and allows users to download community creations. The application is free, open source, and designed to be lightweight and performant.

Aug 22, 2025

Adds streaming audio and lipsync capabilities, introduces Draco compression support for video files, and includes minor fixes and improvements to the TalkingHead library.

May 29, 2025

A hotfix release addressing layer surface handling issues, cursor hover events, renaming urgency actions, and updating the Smithay library to fix various bugs.

May 25, 2025

Introduces the Overview feature for spatial workspace navigation, new screencasting features like dynamic cast targets and windowed fullscreen, and improvements to the screenshot UI.

May 17, 2025

This release introduces new channel concatenated control conditioning trainers for Wan2.1 and CogView4, adds support for FLUX.1-dev and Wan2.1 I2V models, implements multi-provider attention support, and improves regional compilation capabilities.

Apr 25, 2025

The initial stable release adds support for CogView4 and Wan 2.1 models, releases training checkpoints on HuggingFace, re-enables Accelerate as a parallelization backend, and introduces opt-in precomputation for smaller datasets.

Mar 21, 2025

Introduces support for blend shapes based lip-sync animation in the speakAudio function and fixes a bug related to converting numbers starting with zero to words.

Mar 3, 2025

Enhances the bolt.diy editor with support for reasoning content, new dynamic models (DeepSeek, Gemini 2.0), enhanced Code Context features, a new Settings UI V3, and Netlify one-click deployment support.

Feb 28, 2025

The initial alpha release supports diffusion models including CogVideoX T2V, LTX Video, and Hunyuan Video, featuring LoRA and full-rank finetuning capabilities alongside low-bit memory optimizers.

Feb 24, 2025

This release focuses on expanding AI model support and improving the developer experience by adding new providers and enhancing context handling, alongside fixes for Git imports and Docker support.

Jan 22, 2025

A minor hotfix release addressing a bug where the starter template auto-selection failed when no GitHub token was present.

Dec 31, 2024

This release introduces dynamic bones for custom models and enhances the morph target animation system. It also adds audio-reactive head movements and improved mixer controls.

Nov 18, 2024

Minimal release notes: "+ Key transparent image in webapp + Stop inverting black/white for binary images"

Jun 5, 2024

Apr 20, 2024