xAI Imagine Video 1.5: Reference Tools and 1080p AI Video Generation
xAI's Imagine Video 1.5 adds image reference, voice reference, and 1080p output, while a lite variant brings faster, lower-cost generation to Runway Dev.
Reference-Driven Video Generation
On October 19, 2026, xAI released Imagine Video 1.5, an updated version of its video generation model with three major additions: image reference, voice reference, and 1080p output resolution. The release signals xAI's continued investment in multimodal AI capabilities beyond text generation, positioning the company as a competitor in the increasingly crowded AI video market.
Alongside the full model, xAI introduced Imagine Video 1.5 Lite, a faster and lower-cost variant. On October 1, 2026, Runway Dev added this lite version to its developer platform, making it accessible through Runway's unified API infrastructure. The lite model supports 480p, 720p, and 1080p outputs with durations from 1 to 15 seconds for both text-to-video and image-to-video generation.
Image and Voice Reference Features
The image reference capability allows users to supply visual style guides, character references, or scene compositions that the model uses as anchors during generation. Rather than describing a visual style through text prompts, creators can provide actual images that define color grading, lighting conditions, and aesthetic direction. This produces more consistent results across multiple generations and reduces the ambiguity inherent in text-only prompting.
Voice reference enables the model to match generated video content to specific vocal characteristics. For creators producing narrative content, this means characters can maintain consistent vocal identities across multiple clips. The feature integrates with the model's understanding of lip synchronization, improving the coherence between visual and audio elements in generated outputs.
Resolution and Format Options
Imagine Video 1.5 supports multiple output resolutions. Text-to-video accepts width:height ratios across 480p, 720p, and 1080p, with specific dimensions including 848x480, 1280x720, and 1904x1072. Image-to-video uses an auto-resolution system that matches the output to the input image's aspect ratio at the selected quality tier: auto_480p, auto_720p, or auto_1080p.
The full model's 1080p capability places it in direct competition with Kling 4.0 and Google Veo 3.1 for high-definition AI video production. For developers building applications where resolution directly impacts user perceived quality, such as advertising content or cinematic previews, native 1080p output eliminates the quality loss associated with upscaling lower-resolution generations.
Runway Dev Integration
Runway Dev's addition of Imagine Video 1.5 Lite on October 1, 2026, gives developers another model option within Runway's unified ecosystem. The platform now hosts Kling video models, Ideogram image models, Runway's native generation tools, and xAI's video model under a single billing and API management system.
For developers, this consolidation means reduced integration overhead. Instead of maintaining separate API connections, authentication systems, and billing relationships with Kuaishou, xAI, and Runway, a single Runway Dev account provides access to all four model families. The platform's credit system enables unified cost tracking across different model types.
Competitive Position
xAI enters a market where established players have already released multiple generations of video models. Kling 4.0 offers 30-second native generation, 15 multimodal references, and 4K output on its roadmap. Google Veo 3.1 adds native audio generation. Sand.ai's MAGI-2 Preview pursues autoregressive generation with trillion-parameter scale.
xAI's differentiation strategy appears to focus on integration breadth and developer accessibility. By making Imagine Video 1.5 Lite available through Runway Dev at competitive pricing, xAI reduces friction for developers who already use Runway's platform. The reference features address a genuine user need for consistency across generations, a pain point that remains challenging even for leading diffusion-based models.
Whether xAI can capture meaningful market share depends on whether Imagine Video 1.5's quality matches or exceeds competitors at similar price points, and whether Grok's growing user base creates a natural distribution channel for video content. As of mid-October 2026, independent benchmark comparisons remain limited. Pricing for the Lite variant through Runway Dev starts at approximately $0.02 per second of generated video, making it one of the more affordable 1080p options currently available to developers.
Sources: AI D-A-M-N, Releasebot Runway API Updates, Runway Dev Documentation