ChangelogLATEST · v3.1.0

Faster, steadier,
and more capable.

Release notes for LocModel. The latest version, v3.1.0, focuses on multi-GPU, video LoRA, Agent web search and long-run stability.

Multi-GPU ParallelVideo LoRA More controlAuto-restart Crash recovery8 updates Improvements
v3.1.0 LATEST

Faster, steadier,
and more capable.

v3.1.0 focuses on multi-GPU, video creation and long-run stability. It adds video LoRA import and model crash monitoring with auto-restart, and improves Agent web search, video generation speed and multi-channel acceleration.

  1. 01

    Creation

    Image mode fix

    Fixed cases where image mode would not start or generate in some environments.

  2. 02

    Multi-GPU

    Multiple GPUs

    Use multiple GPUs at once, with automatic allocation by VRAM and compute.

  3. 03

    Agent

    Better web search

    Improved Agent web search, information gathering and citations for more complete results.

  4. 04

    Video

    Faster video generation

    Shorter waits and more stable consecutive video generations.

  5. 05

    Video

    Video LoRA import

    Import LoRA models for video generation to control characters, style and motion.

  6. 06

    Stability

    Crash monitoring & auto-restart

    Monitors model processes and restarts automatically after a crash.

  7. 07

    Performance

    Faster multi-channel acceleration

    A faster multi-channel acceleration service for stability and inference speed.

  8. 08

    Compatibility

    More GPUs and tools

    Supports more GPUs, broader tool calling and a range of detail fixes.

Download v3.1.0

Windows is primary · macOS and Linux also available

macOS first launch: No certificate yet. After downloading, right-click → Open. If it still will not install, run this in Terminal: sudo xattr -rd com.apple.quarantine /Applications/LocModel.app

v2.1.0 PREVIOUS

Cross-platform, multilingual,
and easier to use.

This update brings LocModel to macOS and Linux and adds more interface languages. Chat, Agent, image generation, video and model downloads also received targeted improvements.

01Platform & ecosystem

  • Multilingual support Added traditional Chinese, English, Japanese, Korean and other interface languages.
  • In-app updates Added in-app online updates, you can upgrade directly after checking the new version.
  • macOS / Linux builds Added macOS Universal version and Linux version download, Windows is still the main platform.

02Models & compatibility

  • Better third-party APIs Optimize the access process and compatibility of third-party APIs.
  • Better GGUF support Optimize GGUF file recognition and loading, and improve compatibility of different quantization files.
  • Resume download fix Fixed the problem of being unable to continue after downloading is interrupted.

03Chat, Agent & fixes

  • Chat history compression Reduce the context occupied by long conversations and retain key information longer.
  • Better Agent execution Optimize task disassembly, step status and file operation process to make execution more stable.
  • Other fixes Fixed several interface, performance and stability issues.

04Creation

  • Image model list fix Fixed display and loading issues in the drawing model list in some environments.
  • Inference engine update Updated the inference engine and fixed the engine calling problem of raw images and video models in some environments.
  • Better video generation Optimize video generation length and image quality, and improve generation stability.

v2.1.0 is archived. Download the latest v3.1.0 above.