xAI launches Grok Video 1.5 at 86% lower price than Sora
xAI has launched Grok Imagine Video 1.5 into general availability via its developer API at an aggressive price point of $4.20 per minute for 720p output. The model leverages an autoregressive model architecture named Aurora to improve temporal motion coherence, though this sequential design currently limits resolution to 720p. The release intensifies market competition following OpenAI's deprecation of Sora and Google's higher pricing tiers for Veo.
Key Takeaways
- API pricing is set at $0.14 per second for 720p and $0.08 per second for 480p, substantially undercutting Google Veo’s $24 per minute quality tier.
- The Aurora engine generates frames sequentially to ensure temporal coherence, a design choice that currently limits maximum resolution to 720p at 24fps.
- Integrated audio and lip-syncing are generated in a single inference pass, eliminating the need for separate audio tools required by competitors like Runway or Kling.
- Video 1.5 Fast variant reduces 6-second clip generation time to 25 seconds, a 40% improvement in latency for iterative developer workflows.
Why It Matters
The release establishes a new price floor for high-fidelity AI video during a period of competitor retraction, specifically OpenAI’s recent deprecation of the Sora API. By prioritizing temporal stability over raw resolution, xAI is betting that creative directors value consistent camera movement and subject identity more than 1080p pixels for rapid prototyping. This shifts the competitive focus from sheer visual quality to cost-efficiency and workflow integration. For the broader ecosystem, the inclusion of native audio in a single-pass generation simplifies the technical stack for automated content pipelines. Watch for the release of the promised Pro Mode to see if xAI can overcome the architectural scaling limits of its sequential Aurora engine to reach 1080p.
Additional Context
The launch of Grok Imagine Video 1.5 follows a period of intense regulatory scrutiny for xAI. Per TechTimes reporting in March 2026, the company faced federal lawsuits and investigations across the U.S., EU, and UK regarding the generation of non-consensual explicit imagery using earlier versions of the Grok toolkit. In response, xAI restricted generation access to paid tiers in early 2026 and implemented more stringent content classifiers to prevent the misuse of its image-to-video capabilities. These safety measures are now a standard part of the Imagine API stack as xAI seeks to position its tools for enterprise-grade creative applications. Competitive pressure in the AI video space has shifted toward specialized transformer models and utility-based features. In May 2026, Kling and Seedance 2.0 announced enhanced character consistency tools to compete for professional animation workflows, according to industry benchmarks. While Sora 2 was praised for its aesthetic quality, its high compute costs—cited by OpenAI as a factor in its April 2026 discontinuation—created a vacuum that xAI is now aggressively filling with its Aurora architecture. Furthermore, per Reuters in June 2026, the wider streaming industry is increasingly looking toward these low-cost models for localized marketing assets and automated social media promotional clips, where the 720p resolution cap is less of a hinderance than high per-minute generation costs.
Read full article at techtimes.com