LM Studio Beta Releases
🪲 🗣️ Please report bugs and feedback to:
- Email: bugs [at] lmstudio [dot] ai
- Discuss on discord: LM Studio Community Discord Server
- GitHub: lmstudio-ai/lmstudio-bug-tracker
Beta and Experimental Releases
| Version | Build | OS | Arch | Last Updated | Download URL |
|---|---|---|---|---|---|
0.3.14 | 4 | Windows | x86_64 | 2025-03-26 | LM-Studio-0.3.14-4-x64.exe (x86) |
0.3.14 | 4 | Mac | arm64 | 2025-03-26 | LM-Studio-0.3.14-4-arm64.dmg (Apple Silicon) |
0.3.14 | 4 | Linux | x86_64 | 2025-03-26 | LM-Studio-0.3.14-4-x64.AppImage (x86) |
0.3.14 | 4 | Linux | x86_64 | 2025-03-26 | LM-Studio-0.3.14-4-x64.deb (x86) |
0.3.14 | 4 | Windows | arm64 | 2025-03-26 | LM-Studio-0.3.14-4-arm64.exe (ARM64) |
LM Studio is provided under the terms of use.
Release Notes - LM Studio 0.3.14 Build 4 (Beta)
Build 4
- [Advanced GPU controls] Fix bug where disabling a GPU would cause incorrect offloading when > 2 gpus
- [Advanced GPU controls][CUDA] Improved stability of Smart GPU Memory Management mode
- Added GPU controls logging to "Developer Logs"
- Fixed a bug where sometimes editing model config inside the model loader popover does not take effect
Build 3
- [Advanced GPU controls] Fixed a bug where intermediate buffers were being allocated on disabled GPUs
- Fixed "OpenSquareBracket !== CloseStatement" bug with Nemotron model
- Fixed a bug where Nemotron GGUF model metadata was not being read properly
- [Windows] Fixed: Make sure the LM Studio.exe executable is also signed. Should help with anti-virus false positives
Build 2
- Optimized Smart GPU memory management mode in long context situations
- Speculative decoding draft model now respects GPU controls
- [CUDA] Fixed a bug where model would crash with message "Invalid device index"
- [Windows ARM] Fixed chat with document sometimes not working
Build 1
- New: GPU Controls 🎛️
- On multi-GPU setups, customize how models are offloaded onto your GPUs
- Selectively enable/disable individual GPUs
- CUDA-specific features:
- "Fill in Order" mode: Automatically offload models following your specified GPU priority sequence
- "Smart GPU Memory Management" mode: If a model load won't fit in VRAM, use RAM+CPU instead
of allowing shared memory spillage (slow) or risking crashes
- Can disable all GPUs when this mode is enabled, forcing all models to offload to RAM/CPU
- How to open GPU controls:
- Windows:
Ctrl+Shift+H - Mac:
Cmd+Shift+H
- Windows:
- How to open GPU controls in a pop-out window:
- Windows:
Ctrl+Alt+Shift+H - Mac:
Cmd+Option+Shift+H - Benefit: Manage GPU settings while models are loading
- Windows:
- On multi-GPU setups, customize how models are offloaded onto your GPUs
- LG AI EXAONE Deep reasoning model support
- Improved model loader UI in small window sizes
- Improve llama model family tool call reliability through LM Studio SDK and OpenAI compatible streaming API
- [SDK] Added support for GBNF grammar when using structured generation
- [SDK/RESTful API] Added support for specifying presets
- Fixed a bug where sometimes the last couple fragments of a prediction are lost