LM Studio

LM Studio is a graphical desktop application for discovering, downloading, and chatting with open models entirely offline. It includes a built-in server that mimics the OpenAI API for local development and features an agent mode named Bionic for document and code tasks. The application is designed to run local large language models directly on a user machine using backends like llama.cpp and MLX.

• Discover and download local LLMs directly within the app for offline use
• Agent mode named Bionic for creating and editing documents, slides, PDFs, and software
• Real-time voice transcription with audio data processed locally
• Built-in server that mimics the OpenAI API for local development
• Support for running frontier open models such as GLM 5.2, Kimi K3, and DeepSeek V4 Pro via cloud credits or local hardware
• LM Link for connecting up to 5 devices
• Developer tools including JavaScript and Python libraries, a command-line interface, and llms.txt support

Pros

• Operates entirely offline for privacy-focused local model execution and voice transcription
• Provides an OpenAI-compatible local server to facilitate software development
• Offers cloud inference options featuring Zero Data Retention for demanding open models

Cons

• Available in initial preview specifically for macOS and Windows, which may exclude other operating systems
• Web search tool and cloud services require being logged in
• Pricing and plan details for Bionic Pass are marked as coming soon

LM Studio fits developers and users looking to run, test, and build applications with open-source large language models locally on their own machines without sending data to external servers. It is also suited for individuals requiring offline voice transcription and agentic workflows for coding and document creation. It is a poor fit for users seeking cloud-only proprietary models or those operating outside of macOS and Windows during the initial preview phase.

The base product offers a free tier at $0 for running local LLMs and voice transcription on a local machine. A pay-as-you-go cloud credits model is available for US-based inference using frontier open models with Zero Data Retention by default, with distinct per-million-token input, cached input, and output prices listed for models like DeepSeek V4 Flash, DeepSeek V4 Pro, GLM-5.2, Kimi K2.6, Kimi-K2.7-Code, and Kimi K3. Additionally, a Bionic Pass tier has pricing and plan details coming soon.

Unlike pure cloud-based artificial intelligence interfaces, LM Studio operates as a native desktop application focused on offline execution using the LM Studio runtime, llama.cpp, and MLX. It bridges local execution with cloud flexibility by offering optional US-based cloud inference with Zero Data Retention for frontier open models alongside its local features.

macOS, Windows, llama.cpp, MLX, OpenAI API, GLM 5.2, Kimi K3, DeepSeek V4 Pro, DeepSeek V4 Flash, Kimi K2.6, Kimi-K2.7-Code, lmstudio-js, lmstudio-python, and the LM Studio CLI (lms).

LM Studio is worth trying for developers and desktop users who want a graphical interface to manage, test, and run local AI models without sending data to external servers. The free local tier provides strong offline utility, while optional pay-as-you-go cloud credits offer access to heavier frontier models with zero data retention by default. Business deployment requires checking commercial terms, and users needing advanced cloud passes should monitor forthcoming plan details.