Company Dokar, Inc. Has published version 4.40 by Dokar Desktop. An important news doctor is a model runner, with developers larger language model (LLM) can perform locally on the command line. The dock AI agent will also use model reference protocol (MCP) in future and hence it can also be integrated into the workflows that go beyond doctors.
Even without a container: Work locally with AI model
The Docker Model Runner is now integrated as a plug-in in the Dokar Desktop and is activated by default-but initially still in a beta version. According to the declared target Announcement in Dokar BlogDevelopers and developers to open an easy way to work quickly and easily with the AI ​​model locally that outside a container. To do this, Docker can be models of the hub docker model pull Download and save locally.
The execution and interaction with Olama-Commandable with LLMS-Commandable is done through the command line, alternatively into a prompt or chat mode. Dokar has integrated the open source library llama.CPP and has integrated access through Openai API for encroafions. In addition, models are packed as OCI artifacts, so that they can be classified as reliable in registers with installed CI/CD workflows.
During the beta phase, the doors model runner is initially available only for MACOS and uses GPU acceleration of the Apple-Silicon processor on these platforms. The support for Windows and other operating systems should be followed. In addition, doors also want to enable developers to adapt and publish their own AI models in the future.
IX practice workshop: cloud-country software development with cubernets and doorsAI helped beyond Dokar
As an intelligent assistant, the Dokar AI Agent has long been supporting the Dokar AI Agent. With the release of Dokar Desktop 4.40, the agent should achieve even more skills, including downloading and managing local files and executing git operations and shell commands. The dock also offers full support for the AI ​​agent model reference protocol (MCP), through which AI agents and models can be connected to external data and tools. Docker AI agent can act as MCP client and provide its function as MCP server. In this way, support from the agent can also be used directly in Dokar Desktop (GUI and CLI) -other environment such as Cursor IDE or Cloud Desktop.
Other innovations concerned, along with other things, desktop settings launched in version 4.36, which now include compliance reporting for doors trade customers. A complete observation of all changes in version 4.40 Distribute dock blogs as well as Released notes,
(Map)
AMD Ryzen: AI Safety Interval in Software
