New Dev Models feature appears in xAI Grok UI

xAI is working on a new Grok update to introduce a new option labeled “Dev Models.” This discovery came from screenshots shared by an independent app researcher, @nima_owji, revealing early signs of a potential developer-focused tool within the interface.

The screenshot shared on X (formerly Twitter) shows the model selector interface, including new Dev Models. Other tabs like Manage and Docs appear nearby, suggesting integration into existing management tools.

When users select Dev Models, the interface currently shows an error message: “Failed to load dev models,” accompanied by a simple Retry button. This indicates the feature remains in an early testing phase and is not yet available for general use.

A second screenshot provides more insight into this feature. It shows a draft screen for creating a new model. The layout includes input fields designed for customization. Users can enter a name to display in the user interface, add a description, and select a base scheduler configuration from a repository dropdown, which appears while loading available options.

xAI Grok dev models

Another field allows specification of a model address using either gRPC or HTTP endpoints, pointing to technical integration capabilities. Additional sections cover system prompts, where core behavioral instructions are defined, and developer prompts for more specialized guidance.

These elements suggest the feature could enable developers to connect external AI models or build tailored versions directly within Grok. The presence of endpoint support hints at compatibility with remote servers, potentially opening doors for advanced integrations. However, the screen also displays a “Failed to load models” error, reminding us that the tool is still under development.

This might serve as an internal testing ground for upcoming xAI models, allowing team members to experiment privately. Alternatively, it could evolve into a public feature supporting community-created models, similar to how other platforms allow user-shared customizations. Such a feature would let individuals design and deploy specialized AI variants.

While details about dev models remain limited and the feature shows loading issues, this development aligns with xAI’s ongoing Grok AI improvement scope to more open and customizable AI experiences.