Grok Imagine Image 2.0 now available on Vercel AI Gateway

Vercel’s AI Gateway has integrated xAI’s Grok Imagine Image 2.0 Preview, a model designed to interpret detailed instructions while maintaining typography and layout integrity in complex visuals such as infographics, posters, and title screens. The model ensures small text remains legible and preserves structural coherence in dense compositions. Users can access the preview via imagine.vercel.sh, where the model runs on Vercel’s AI Gateway infrastructure. To deploy the model, developers must specify xai/grok-imagine-image-2.0-preview as the model identifier and invoke the generateImage function from the AI SDK.

The model introduces support for image editing, allowing users to modify specific elements of an image while retaining unchanged details across subsequent generations. Edits are applied by including an input image in the prompt.images field alongside the instruction, enabling targeted adjustments without altering unrelated components. This feature aims to streamline iterative design workflows by reducing the need for repeated manual refinements. The editing capability is designed to maintain subject consistency, ensuring coherence in multi-step creative processes.

Developers can select output resolution tiers of 1k or 2k by configuring providerOptions.xai within the API call, with an optional parameter n to generate multiple images per request. These settings provide flexibility in balancing quality and computational cost based on project requirements. The resolution options are explicitly defined to align with common use cases in visual content creation, from rapid prototyping to high-fidelity production.

Vercel’s AI Gateway aggregates multiple image generation models, and Grok Imagine Image 2.0 Preview is now included in the supported catalog accessible via the platform’s documentation. The integration follows Vercel’s policy of expanding its AI model ecosystem to include emerging tools while maintaining compatibility with existing workflows. Users can reference the full list of supported models on Vercel’s official documentation page for further details on available options and usage guidelines.