xAI has upgraded Grok Imagine Video, its AI video generator, with native 1080p output, text-to-video generation, and new reference tools that let creators lock a character’s face and voice across multiple shots, the company announced August 1.
The update expands Grok Imagine Video, which launched earlier this summer with image-to-video clips capped at 720p. According to xAI, the tool now accepts up to seven image references per generation — covering faces, objects, or locations — so the same character or setting stays consistent across a sequence of shots. A companion voice-reference feature keeps a character’s voice matched to their face throughout a scene.
Text-to-video is now generally available on grok.com/imagine, xAI’s iOS and Android apps, and its developer API under the model name grok-imagine-video-1.5. The image- and voice-reference features are rolling out over several days, starting with subscribers to xAI’s SuperGrok Heavy and SuperGrok Plus tiers in the US before reaching other plans and regions. Developers can already call image references through the API; voice references require a direct request to xAI.
Racing rivals on video
The release adds to a crowded field of AI video tools. Grok Imagine Video briefly topped an image-to-video leaderboard after its debut, competing against Google’s Veo models and China’s Seedance; OpenAI, meanwhile, pulled its Sora app from some markets earlier this year, citing an unclear business model. xAI has not disclosed specific pricing for the new 1080p and reference features, though it has generally priced its video API below rival offerings.
The upgrade follows xAI’s push to bundle Grok Imagine into its consumer apps alongside Grok 4.5, the chatbot model xAI released earlier this year, as the company competes for both developers and everyday users of AI video tools. For background on the company behind the model, see what xAI is and how it fits into Elon Musk’s businesses.