Docker: Docker Model Runner General Availability

Sep 18, 2025

—

Source URL: https://www.docker.com/blog/announcing-docker-model-runner-ga/
Source: Docker
Title: Docker Model Runner General Availability

Feedly Summary: We’re excited to share that Docker Model Runner is now generally available (GA)! In April 2025, Docker introduced the first Beta release of Docker Model Runner, making it easy to manage, run, and distribute local AI models (specifically LLMs). Though only a short time has passed since then, the product has evolved rapidly, with continuous…

AI Summary and Description: Yes

Summary: Docker Model Runner (DMR) is a new product from Docker aimed at simplifying the management and deployment of large language models (LLMs) for developers. Its integration into the existing Docker ecosystem allows for secure and efficient workflows, making it easier for developers to build AI applications without needing to adopt new tools.

Detailed Description:
Docker Model Runner (DMR) is a noteworthy advancement for developers looking to work with AI, particularly in the realm of Large Language Models (LLMs). The product has recently transitioned from a beta phase to general availability, reflecting significant improvements in its capabilities and reliability. Here’s a comprehensive overview of its features, implications for security and compliance, and future directions:

– **Developer-Centric Design**: DMR is designed specifically for developers, allowing them to pull and distribute AI models using familiar Docker commands and tools.

– **Integration with Docker Ecosystem**:
– Works seamlessly with Docker Desktop and Docker Engine.
– Compatible with OpenAI APIs and enables packaging of models in OCI-compliant formats.

– **Security and Compliance**:
– Runs within existing enterprise security and compliance frameworks, minimizing disruption to established workflows.
– Features like Registry Access Management (RAM) promote policy-based access controls, which are crucial for maintaining security within scalable environments.

– **Highlighted Features**:
1. **GPU Acceleration**: Supports various hardware configurations (macOS with Apple Silicon, NVIDIA on Windows) to enhance performance.
2. **CLI and UI Accessibility**: Offers both command-line and graphical interfaces, facilitating easier use for first-time AI developers.
3. **Open Source**: Being free and open source lowers barriers for developers and promotes community collaboration.
4. **Sandboxing**: Runs in a controlled environment to ensure isolation from user data and main system interference.
5. **Flexible Model Distribution**: Supports pulling from Docker Hub and HuggingFace, accommodating various model formats (OCI & GGUF).

– **Future Enhancements**:
– Plans for continuous improvement in user experience and performance, including streamlined onboarding processes for new users.
– Intent to expand the model catalog available to developers on Docker Hub, ensuring access to the latest AI models as they become available.

– **Practical Implications**:
– For security and compliance professionals, DMR represents a new tool that can be integrated into existing workflows without major alterations to security policies.
– The product’s emphasis on compliance and security should ease concerns for organizations wary of adopting new technologies.

The Docker Model Runner is a considerable development for both developers and compliance professionals, creating a dedicated platform that streamlines AI model deployment while maintaining a strong focus on security and ease of use. As the AI landscape continues to evolve, tools like DMR will play a pivotal role in bridging the gap between innovation and secure practice in AI development.

1 2 2025 3 4 5 a acceleration access access control access controls access management accessibility Act advancement age AGI AI AI applications AI developers AI development AI landscape ai model AI models All allow alt and API APIs app Apple Apple Silicon Application applications art as at ated availability based based Access Control based access controls being Bi board bot Box C capabilities catalog centric centric design CERN CI CIA cli co Col collaboration command community community collaboration compliance compliance framework compliance frameworks compliance professionals concerns Configuration configurations continuous continuous improvement control controlled environment controls D data de deployment design Desktop developer developers development disruption Docker Docker Desktop Docker Engine Docker Hub Docker Model Runner e ease of use ecosystem efficient enterprise Enterprise Security environment environments ERP exp experience face feature features first for framework frameworks free future future directions g Gen general GIS GPU graph H hardware hardware co hardware configuration hardware configurations high Highlight http HTTPS hugging Huggingface ICO implications implications for security in innovation integration intent inter interface Interfaces io Iron IRS isolation ite J k l Labor land language language model language models large large language model large language models Large Language Models (LLMs) led Li liability line llm llms lm local low M mac MacOS making man management mini ML Mode model model catalog model deployment model distribution Model Runner models N new no Nvidia o of off on onboarding onboarding process one only ons open open-source openai OPM opt organization organizations oS out over packaging per performance phi platform play policies policy policy-based access control practical implications pre pro process processes product professionals ps R rate RCE re real Registry Registry Access Management release reliability Ro Role s sandbox sandboxing scalable sec secure security security and compliance security policies SHA short side Sig Silicon Sim source specific SSE SSL support system T tech technologies ted test the Time to tool tools TP transition UI up US use user user data user experience Users V Ware Wi Wind Windows workflow workflows x z