xAI completes 'Grok Imagine,' pushing Grok into multimodal image and video generation

Elon Musk posted 'Done with Grok Imagine' on July 5, signaling the completion of xAI's image- and video-generation feature — a milestone in xAI's push to make Grok a full multimodal assistant rather than a text-and-voice chatbot. Grok Imagine has been in development since August 2025 and spent months in beta; it is powered by xAI's Aurora image model and supports a range of aspect ratios and visual styles.

The launch slots into a broader xAI product cadence: the company recently moved its no-code Voice Agent Builder to general availability at $0.05 per minute with 80+ voices and voice cloning, and Musk has flagged a 1.5-trillion-parameter Grok model for July and a 2-trillion-parameter model in August. Together they show xAI racing to match OpenAI, Google and Anthropic across modalities and agentic tooling simultaneously.

Competitively, Grok Imagine enters a crowded field — Google's Nano Banana image models and Gemini Omni Flash video, OpenAI's image tools, and a vibrant open-source scene around Krea and Ideogram. xAI's differentiator is tight integration with X's real-time firehose and Grok's less-filtered posture, which is both a draw and a moderation risk.

Caveats: Musk's 'done' announcements have historically preceded staggered rollouts, and image/video generation invites the same deepfake and copyright scrutiny competitors face. What to watch: pricing, rate limits, and whether Grok Imagine ships inside the X app broadly or behind a Premium+ paywall.