Elon Musk says 'Grok Imagine' image and video generation is complete

Elon Musk announced that 'Grok Imagine,' xAI's image and video generation capability, is development-complete, following up on July 7 with a video preview showing new or improved generation features. The tool uses xAI's proprietary Aurora model to turn text prompts into images and video, extending Grok beyond text into full multimodal generation.

The move lands in a crowded week for image generation: Meta launched Muse Image (ranked #2 in the Image Arena) and ByteDance's Seedance 2.5 video model — supporting up to 50 references and 30-second 4K video — may launch imminently. Video generation specifically has become the new competitive frontier, and xAI joining raises the stakes against OpenAI, Google's Veo lineage and the Chinese video models.

Strategically, this is part of the broader SpaceXAI push — the rebrand, the Grok 4.5/Cursor model reportedly shipping this week, and a Voice Agent Builder — as the company tries to build a full multimodal, agentic stack rapidly post-acquisition. Bundling image/video generation into Grok, distributed through X, gives xAI a consumer-creation channel comparable to Meta's Instagram/WhatsApp integration.

The caveats are familiar with xAI announcements: 'development complete' is not the same as broadly available, quality claims rest on a single video preview, and xAI's Aurora has historically trailed leading image models on fidelity and safety. Content-moderation questions loom large given Grok's looser guardrails and video generation's abuse potential. Watch for a public benchmark or Arena entry to gauge whether Grok Imagine is competitive or a catch-up release, and whether video generation ships to all X users or a premium tier.