Elon Musk's video AI just jumped to the top of the leaderboard. It's also $25.80 cheaper per minute than the tool it dethroned. That combination should worry every other AI video company on the planet — and it should make you ask one question before you touch it.
Grok Imagine Video 1.5 went fully live on June 16, 2026, and it didn't tiptoe into the market. It walked straight to the top of the Image-to-Video Arena leaderboard — the closest thing the AI video world has to an objective scoreboard — with an Elo score sitting around 1,330. That's enough to beat Sora 2, Google's Veo 3.1, Seedance 2.0, and Kling in blind side-by-side testing.
I dug into the pricing, the architecture, and — because nobody else seems to want to say it out loud — the platform's actual safety record. Here's the full picture.
The Number That Matters: $4.20 a Minute
The price is what makes this genuinely interesting, not just another model topping a leaderboard for a week before getting dethroned itself.
| Feature | Grok Imagine Video 1.5 | Sora 2 Pro (pre-deprecation) |
|---|---|---|
| Price (720p, per minute) | $4.20 | $30.00 |
| API price per second | $0.08 (480p) / $0.14 (720p) | N/A — discontinued |
| Max clip length | 15 seconds | Varies by tier |
| Audio generation | Native, same generation pass | Separate process |
| Architecture | Aurora (autoregressive) | Diffusion-based |
| Fast-mode generation time | ~25 sec for a 6-sec clip | 40+ sec (prior Grok version) |
| Leaderboard position | #1 (Image-to-Video Arena) | Discontinued April 2026 |
For API users, that breaks down to $0.08 per second at 480p and $0.14 per second at 720p. xAI also runs a consumer-facing version at grok.com/imagine, where a free tier gives you 5 credits a day to test the waters before paying for anything.
💡 Why this timing isn't an accident: OpenAI shut down the standalone Sora 2 app on April 26, 2026. The Sora 2 API is scheduled to sunset entirely on September 24, 2026. xAI didn't just release a competitive product — it released one six weeks after its biggest rival started walking out the door.
Think about what that price actually buys. A batch of ten 6-second social clips at 720p runs roughly $4.20 total through the API. Under the old Sora 2 Pro pricing, that same batch would have cost closer to $30. That's not a minor discount — it's the difference between "we can't justify AI video in the budget this quarter" and "let's just try it on three campaigns this month and see what sticks."
What It Actually Does
Strip away the marketing, and Grok Imagine Video 1.5 does one specific thing: it takes a still image plus a text prompt describing motion, and turns that into a video clip up to 15 seconds long, at 480p or 720p, with synchronized audio baked into the same generation pass. No separate audio step. No syncing it up afterward in an editor.
That last part matters more than it sounds. Most competing tools generate silent video first and bolt audio on as a second process. Grok does both at once, which is part of why the output feels more cohesive instead of like two AI systems that never talked to each other.
There are four animation presets — Normal, Fun, Custom, and Spicy — which set the overall tone of the generated clip. More on why that last one matters in a minute.
The Engine Under the Hood (And Why It's Both a Strength and a Limit)
This is the part most coverage skips, and it's the part that actually explains what you're getting.
Grok Imagine runs on Aurora, xAI's autoregressive video architecture. In plain English: instead of generating an entire video clip at once the way diffusion-based models do, Aurora builds it frame by frame, in sequence, with each new frame conditioned on every frame that came before it.
That's exactly why the camera movements look stable and why subjects don't randomly drift or warp between frames — the model is, quite literally, checking its work against everything it already generated before adding the next piece.
It's also exactly why this thing caps out at 720p for now.
💡 The tradeoff nobody puts in a headline: Scaling from 720p to 1080p means roughly 2.25 times more pixel data per frame. In a diffusion model, that extra load gets spread across parallel processing. In an autoregressive model like Aurora, every one of those extra tokens has to be processed in sequence — there's no shortcut. That's a real architectural ceiling, not a "coming soon" feature gap.
What you do get in exchange: genuinely fast turnaround. Fast Mode generates a 6-second, 720p clip in about 25 seconds — down from over 40 seconds in the previous version. If you're an agency cranking out dozens of social clips a day, that difference adds up fast.
Who This Is Actually Built For
I'll be blunt: if you're a Hollywood production house chasing 4K cinema-grade output, this isn't your tool yet. The resolution ceiling rules that out.
But that's not who's going to use this in volume.
✅ Grok Imagine Video 1.5 is best for: Social media managers who need a constant stream of short, attention-grabbing video on a tight content budget. Freelancers and small agencies offering "AI-assisted video" as a service — at $4.20 a minute, client retainer margins suddenly look a lot better. E-commerce sellers who want product demo clips without booking a studio. Developers building video generation into their own apps, where per-second billing means you only pay for what actually ships.
If you've been priced out of AI video tools because Sora 2 Pro's $30-a-minute rate made the math impossible for anything beyond a single hero shot, this is the first serious crack in that wall.
The Part the Hype Cycle Keeps Leaving Out
Here's where I have to slow down, because being straight with you matters more than keeping the energy up.
⚠️ xAI's Imagine product line has a documented safety problem. Going back to its original August 2025 launch, outlets including The Verge, CNN, and NBC News have reported that the tool generated sexualized, non-consensual imagery of real people — including public figures — often with minimal or no explicit prompting required. The issue drew regulatory attention on multiple continents. The UK's media regulator made formal contact with xAI over what it called "very serious concerns." California's Attorney General opened a formal investigation and issued a cease-and-desist letter in January 2026. Indonesia and Malaysia temporarily restricted access to the platform over the same concerns before partially lifting those restrictions in early 2026.
xAI's own acceptable use policy explicitly bans non-consensual depictions of real people and states the company reports suspected child exploitation material to the National Center for Missing and Exploited Children. The company has since added age-verification gates and geographic restrictions in response to the pressure, and access to the Spicy preset now generally requires a paid subscription and an 18+ confirmation in the US.
Why bring this up in an article about video generation speed and pricing? Because if you're a brand, a freelancer, or an agency planning to build a client-facing workflow around this tool, you need to know exactly what reputational and legal exposure sits next to the feature set — before a client asks you about it, not after.
The video generation technology itself — the Aurora engine, the pricing, the speed — is genuinely impressive, and it's real news. The safety track record sitting right next to it is also real, and it isn't going away just because the leaderboard score is good.
So, Is It Worth Using?
For the actual job most people will use it for — fast, cheap, decent-quality social and marketing video — yes, with eyes open. The $4.20-per-minute price point alone changes what's financially viable for small creators and agencies that got priced out of this category entirely under the old Sora pricing.
For anything brand-sensitive, client-facing, or involving real people's likenesses, build in a manual review step before anything ships. That's not optional caution — it's the difference between using a powerful tool responsibly and becoming someone else's cautionary headline.
The bottom line: xAI built something technically excellent and made it disruptively cheap. It also built it inside a product family with a real, ongoing safety reckoning. Both of those facts are true at the same time, and anyone telling you only one half of that story isn't giving you the full picture.
Frequently Asked Questions
Is Grok Imagine Video 1.5 free?
There's a free tier on grok.com/imagine offering 5 credits a day, enough to test the model before committing to a paid plan. Production use through the API or ImagineArt's Pro plan requires payment.
Can it generate video from text alone, with no starting image?
The current preview model is built around image-to-video generation — you provide a still image plus a motion prompt. Pure text-to-video support is more limited on this specific model version, so check the current model page before committing a workflow to it.
Why is it cheaper than Sora if it's ranked higher?
Partly architecture, partly timing. Aurora's autoregressive design is computationally efficient at 720p, and xAI launched this pricing right as its biggest competitor was winding down its own consumer product — an aggressive, deliberate undercut rather than a coincidence.
Is it safe to use for client or brand work?
The core video technology is solid. The platform's broader safety record around its Spicy preset and non-consensual imagery is a real, documented concern that any agency or brand should factor into their approval workflow — not a reason to avoid the tool outright, but a reason to add human review before anything ships.
How does it compare to Google Veo and Kling?
On the Image-to-Video Arena leaderboard, Grok Imagine Video 1.5 currently ranks above Veo 3.1, Seedance 2.0, and Kling in blind Elo-style testing. Price-wise, it also undercuts Veo, making it the cheapest top-three option as of June 2026.
Last updated: June 19, 2026. Pricing and benchmark data referenced in this article reflect figures published by xAI, the Image-to-Video Arena leaderboard, and reporting from TechTimes as of June 2026. Regulatory details are based on reporting from CNN, NBC News, and other outlets covering xAI's Imagine product line.
0 Comments