v1.0.4
- Fixed Windows update failures: resolved update package verification and execution failures under Windows, ensuring smooth automatic in-app updates.
- Configurable transcription concurrency: added a concurrency setting for local AI transcription to maximize multi-core CPU and GPU performance.
- Refined model settings and expanded model library: redesigned the AI model settings UI and added support for more LLMs and speech-to-text models.
- Disabled Bilibili playlist downloads on Windows: turned off Bilibili multi-part collection downloads in the Windows edition to avoid high resource spikes and queue errors.
- Added Select All in Windows files view: added a Select All button to the Windows files page for faster batch operations and file management.
- Upgraded AI smart analysis UX with video playback: refreshed the AI analysis workflow with integrated inline video playback and preview.
- Cookie-free Douyin downloads: updated the Douyin download engine to fetch watermark-free videos directly without requiring manual cookie export or configuration.
- Added sleep prevention for remote downloads: prevents system sleep during tasks and standby, enabling continuous remote downloads anytime via WeChat.
- Replaced system player with WebKit engine: migrated media preview to a modern WebKit-based playback component for better compatibility and playback smoothness.
- Optimized network proxy display: improved the visual layout and status indicators for proxy configurations, making connection states clear at a glance.
- Turned off default Bilibili collection downloads: changed Bilibili playlist downloading from default-on to opt-in, preventing unintended queueing of large series.
- Streamlined YouTube settings: removed seldom-used advanced options and clutter from YouTube configuration for a cleaner setup experience.
- Improved browser extension display: optimized the presentation and installation entry points for companion browser extensions and cookie export tools.
- Added media transcoding feature: built-in converter to easily transcode downloaded media into universal formats such as MP4 and MP3.
- Enhanced failure diagnostics with detailed error reasons: overhauled failure prompts to display human-readable error summaries and direct access to underlying failure logs.
- Removed blocking model dependency checks: eliminated strict dependency pre-checks, speeding up app launch and model switching.
- Optimized MCP experience with LAN access: enhanced Model Context Protocol (MCP) tooling and enabled local network (LAN) access for multi-device AI agent integrations.
- Added WeChat Channels image downloads: expanded WeChat Channels download support to capture full-resolution images and photo galleries.