Grok Imagine Video 1.5 Tops the AI Video Leaderboard at $4.20 a Minute

A minute of AI video from Sora 2 Pro costs $30. A minute from xAI’s Grok Imagine Video 1.5, released on June 17, costs $4.20. That 86% price gap would be notable on its own. Combined with the fact that Grok Imagine Video 1.5 just claimed the #1 position on the Image-to-Video Arena leaderboard with a 1,473 Elo rating, beating Sora 2, Veo 3.1, Seedance 2.0, and Kling 3.0 in blind user voting, it reshapes the production budget for every creator making short-form video.

What Landed on June 17

Grok Imagine Video 1.5 runs on xAI’s Aurora autoregressive architecture and is available now on grok.com/imagine, through the Imagine API, and on iOS and Android apps. It generates video from text prompts, still images, or a combination of both.

The 52-point Elo jump from Version 1.0 (the largest single-version improvement in the Arena’s history) came from improvements across the board: sharper physics simulation, more realistic fabric and liquid rendering, better lighting consistency across frames, and substantially improved lip-sync accuracy. The Arena measures these through blind user voting on motion quality, prompt adherence, and visual consistency, so the ranking reflects what real users prefer when they cannot see the model name.

Generation speed roughly doubled. A 6-second 720p clip now renders in about 25 seconds, down from 40+ seconds in the previous version. For creators iterating on concepts, that speed means generating five variations in the time one Sora clip takes to finish.

The standout feature is native synchronized audio. Sound effects, ambient noise, and lip-synced dialogue all generate in the same forward pass as the video. No separate audio tool. No post-production sync. Describe a coffee shop scene and you get cups clinking, conversation murmur, and a door chime. Generate a talking head and the mouth movements track the speech. Most competing tools, including Runway Gen-4 and Seedance 2.0, output silent video and leave audio to a separate workflow. Grok eliminates that step entirely.

Four Creative Modes

Grok Imagine Video 1.5 ships with four creative modes that control the visual style of every generation:

Normal produces realistic, naturalistic footage with balanced physics and lighting. Use it for product demos, talking-head content, and anything that needs to feel grounded. A prompt like “woman unboxing a product on a desk with soft window light” renders exactly as described.

Fun exaggerates motion and energy for a more animated, playful visual style. Social creators making meme content, reaction clips, or anything that benefits from visual punch will get better results here than trying to engineer playfulness through prompt wording alone.

Spicy applies dramatic lighting, saturated color grading, and cinematic motion. The visual language reads like a film trailer: high contrast, bold color, and camera movement that feels intentional. Useful for scroll-stopping content on platforms where the first half second determines whether someone keeps watching.

Custom accepts a natural-language description of the style you want. “1970s Super 8 home video,” “corporate keynote stage with LED wall,” or “documentary interview, shallow depth of field, natural light” all work as inputs. This mode is where experienced prompt writers will find the most creative range.

All four modes generate synchronized audio automatically. The audio layer adapts to the scene content: street traffic for urban footage, wind for outdoor shots, ambient chatter for crowd scenes. xAI has not published the exact audio model architecture, but the output quality represents a clear step beyond what was available in any consumer-grade video generator earlier this year.

Camera movement follows natural-language instructions. Ask for a dolly push, a tracking shot, a crane rise, or a slow zoom and the model interprets the direction. This kind of cinematographic control was previously limited to higher-priced or more complex tools like Runway Gen-4 and Luma Ray3.2’s keyframe system.

Aspect ratios cover every distribution channel creators use: 16:9 for YouTube, 9:16 for Reels and TikTok, 4:3 and 3:4 for legacy formats, 2:3 and 3:2 for Pinterest and editorial use, and 1:1 for Instagram grid posts. Switching aspect ratio does not require a new prompt; the model recomposes the scene to fit.

What It Costs (and What the Competition Charges)

For a creator producing 100 minutes of AI video per month (a realistic volume for someone generating social clips, ad variations, or faceless YouTube content), the monthly bill is $420 versus $3,000 on Sora 2 Pro. Annual savings: roughly $30,960.

Access runs through X and xAI subscription tiers:

• X Premium ($8/month): Basic Grok access with limited image and video credits. An entry point for creators already paying for the verified checkmark and monetization features.
• SuperGrok Lite ($10/month): Expanded prompt limits and generation credits at the lowest dedicated Grok price point.
• SuperGrok ($30/month): The creator sweet spot. Unlimited image generation, daily video renders, 128K context window, DeepSearch, and Big Brain mode. One month of full access costs the same as a single minute of Sora 2 Pro video.
• X Premium+ ($40/month): Same Grok 4 model access as SuperGrok, plus enhanced X platform perks including ad-free browsing, larger creator revenue share, longer post limits, and priority reply ranking.
• SuperGrok Heavy ($300/month): Maximum limits across all features, four parallel AI agents for batch workflows. Built for agencies and teams processing video at scale.

Developers building automated pipelines can access the Imagine API directly, where pricing runs on a per-clip credit basis that scales by resolution and duration.

Where It Falls Short

Every pricing advantage has a specification trade-off. Grok Imagine Video 1.5’s limits are clearly defined:

Resolution maxes at 720p. No 1080p, no 4K. For YouTube Shorts (which compress aggressively anyway), TikTok, and Instagram Reels, 720p is adequate. For YouTube longform content, client deliverables, or anything viewed on a large screen, it will not meet the standard.

Clips cap at 15 seconds per generation. Longer sequences require multiple “Extend from Frame” passes, which can introduce visual drift: subtle shifts in lighting, character appearance, or background consistency that compound across segments. Planning shorter, self-contained clips produces better results than chaining long sequences.

Frame rate is locked at 24fps. Sufficient for cinematic content and most social video, but there is no 60fps option for sports, gaming, or high-motion footage.

Style control operates through the four-mode system rather than per-frame keyframes. Creators who need to direct individual frames, adjusting camera angle or subject position at specific timestamps, will find this coarser than the timeline controls available in Luma Ray3.2 or Runway Gen-4.

These constraints define the tool’s sweet spot: high-volume, fast-iteration, cost-efficient generation for social and marketing content. Grok is optimized for generating many clips quickly and selecting the best, rather than crafting a single perfect shot.

Which Creators Should Care

Social-first creators producing daily Reels, TikToks, and Shorts gain the most from this launch. Generate 10 variations of a concept with synchronized audio already attached, pick the strongest, and publish. The Fun and Spicy modes add visual variety that would otherwise require separate color grading or effects work in CapCut or DaVinci Resolve.

Marketing creators and small agencies running product demos, testimonials, or ad variations can feed product photography directly into the image-to-video pipeline. The cost structure makes A/B testing every creative direction practical rather than aspirational. Ten variations at $4.20 each cost $42; the same experiment on Sora 2 Pro costs $300.

Creators already paying for X. If you subscribe to SuperGrok for the AI assistant, research tools, or platform features, video generation is now part of that same subscription. You do not need a separate video tool account for short-form content.

For longform YouTube creators, the math is different. Runway Gen-4 and Kling 3.0 still deliver higher resolution and longer generation windows. The practical approach: prototype and storyboard in Grok where iteration is cheap, then render final assets in a higher-resolution tool once the creative direction is locked.

The AI video generator market in 2026 has followed a consistent pattern, with each new model delivering better visual quality at a premium price point. Grok Imagine Video 1.5 breaks that pattern. The top-ranked model on the public leaderboard is also the cheapest by a factor of seven. For creators managing production budgets, the decision just got measurably simpler.