Grok Imagine Releases a 7-Minute AI 'Odyssey' Episode After Musk Promised a Full Film

TL;DR

On October 23, xAI's Grok Imagine account posted "Episode 1 of our Odyssey series": 7 minutes 12 seconds of AI-generated video served at 3840x2160, credited as a creative collaboration with Wonder Studios and director Dev Walker. Three months earlier Elon Musk promised that Grok Imagine would make a "full-length," historically accurate movie of The Odyssey before the end of 2026. What shipped is a serialized pilot, not a film, and critics at TechCrunch and The A.V. Club called it "completely airless" and "a choppy trailer or sizzle reel." Christopher Nolan's The Odyssey, the live-action film Musk set out to correct, has grossed $1.79 billion worldwide on a $250 million budget. If you make video with generative models, the useful part is the production math: a 15-second-per-clip model, stitched into seven minutes, by a funded studio with a human director.

What shipped

The Grok Imagine post went up at 19:23 UTC on Friday, October 23. The full text: "Watch Episode 1 of our Odyssey series made with Grok Imagine. Let us know in the comments if you'd like a full breakdown of how it was done! Creative collaboration with @WonderStudiosX and director Dev Walker @glumlot_". The attached video runs 432 seconds and X serves it up to 3840x2160, so the 4K claim holds at the file level. By Monday morning UTC the post had about 19,800 likes.

Wonder Studios reposted it two hours later with the line "Homer didn't have Grok… we do." Musk quote-posted it the next morning with four words, "Odyssey with Grok Imagine," and nothing about the feature film. There was no xAI blog post or release note, and nothing about an episode count or a schedule. Homer did not have Grok. He also did not have a December 31 deadline.

The promise, dated

Nolan's film opened worldwide on July 17 (international markets from July 15) to a $264.1 million global opening weekend, $124.5 million of it in the US and Canada. Musk had spent months attacking the casting, in particular Lupita Nyong'o as Helen of Troy, and had called Nolan "an anti-White racist" on X. On July 22 he quote-posted a roughly three-minute Grok Imagine clip by the studio Heavy Pulp and wrote: "Before this year ends, Grok Imagine will make a full-length movie of The Odyssey," adding that it would be historically accurate and true to Homer (Khaleej Times, Far Out). When another user suggested giving Mel Gibson $100 million to make an accurate version instead, Musk replied "I'm down."

Ninety-three days later, the delivery is seven minutes, labeled Episode 1. Variety's Payton Turkeltaub wrote that Musk had "partially followed through." Director Scott Derrickson (Doctor Strange, The Black Phone) was blunter on X, saying Musk's "opinions on art are worthless."

What is in the seven minutes

Per The A.V. Club's Olivia Singh, the episode opens on a brooding Odysseus sharpening a knife and cutting himself. Penelope tells him he has to leave and hands him a small wooden horse to bring home to Telemachus. He and his men sail for Troy, battle footage is intercut with Telemachus growing up and training, and that is the episode. Singh's verdict: "as uninspired as expected," playing "less like the start of a cohesive feature film and more like a choppy trailer or sizzle reel, at best."

TechCrunch's Anthony Ha found the footage better than the AI clips Musk usually reposts and still called it "completely airless": a British-accented narrator carries a long backstory voiceover, bearded men shout stilted lines, and nobody speaks ancient Greek, an odd omission for a project sold on historical accuracy. Ha's closing line: "It's hard to imagine being interested in another seven minutes of this, much less a full movie's worth."

Two reviews are not a consensus, and neither critic was ever going to love this project. Neither one said the pixels were bad, though. Both said the episode gives you no one to care about, and a video model does not fix that for you.

How you get seven minutes out of a 15-second model

Grok Imagine's video model has hard limits that shape what a project like this can be. xAI's developer docs list the current model, grok-imagine-video-1.5, at "up to 15s" per generation. It supports text-to-video, image-to-video, and reference-to-video with up to 14 reference images, 3 voice references, and control over the first, last, and mid-video frames. An extension mode continues a clip from its last frame and merges the two. A separate grok-imagine-video model edits existing footage from a text prompt.

So a seven-minute episode is, at minimum, around 30 clips at the 15-second ceiling, and in practice far more, because real edits cut faster than that and most generations get thrown away. The reference-image and voice-reference features are what keep Odysseus looking and sounding like the same man from shot to shot. Keeping him consistent is most of the craft. It is less like shooting a movie and more like writing a novel one postcard at a time: every fragment has to stand on its own, and the job is hiding the seams.

The money side is public too. xAI's model page prices video by the second, with resolution affecting the total: $0.080 per second for grok-imagine-video-1.5, $0.020 for the lite variant, and $0.050 for the edit model. Multiply the 432 seconds that made the final cut by those rates and you get between $8.65 and $34.58, if every second had been generated exactly once. Nobody generates once. The real bill is the takes you throw away and the people who decide which ones to keep.

The humans in this case are Wonder Studios, an entertainment company founded in 2025 by filmmaker Justin Hackney and marketplace entrepreneur Xavier Collins, with its team spread across London, LA, New York, and San Francisco. Its site says it has raised more than $15 million from LocalGlobe, Atomico, and Adobe Ventures plus angels from OpenAI, Google DeepMind, and ElevenLabs, employs 30-plus people, and claims a 1.5 million-plus audience for season one of its YouTube anthology Beyond the Loop. Wonder credits a director, Dev Walker, a detail Musk's July post did not include: the "AI will make a movie" pitch became "a studio will make a series with the AI."

The scoreboard Musk picked

Musk framed this as a correction of Nolan's film, so Nolan's film is the comparison. Per The Numbers, through October 25 The Odyssey has taken $625.4 million domestic and $1,169.0 million international for $1,794.4 million worldwide, against a $250 million production budget. It runs 172 minutes, is rated R, and is Nolan's highest-grossing film.

Against that, Episode 1 is 7.2 minutes, about 4% of Nolan's runtime, and the three-minute Heavy Pulp clip that started all this was under 2%. "Full-length" has no official definition, but if Musk means anything near Nolan's length, there are roughly 165 minutes left to make in 69 days.

Caveats

• Nobody has published the process. The post offers a "full breakdown" if commenters ask. Until that appears, clip counts, generation counts, how much was Grok Imagine versus other tools, and how much compositing and sound work happened outside the model are unknown.
• "Series" is doing a lot of work. There is no announced episode count, cadence, or statement that the episodes will add up to the promised feature. Neither xAI nor Musk has addressed the year-end deadline since the episode went up.
• Two reviews, both from outlets that cover Musk critically. They agree on what happens on screen; the verdicts are theirs.

What to take from it if you build with video models

• The clip ceiling defines the form. At 15 seconds per generation, a long piece is an editing and continuity project before it is a prompting project. Budget your time there.
• Reference images, voice references, and frame anchors are the consistency toolkit. Learn them before you learn prompt phrasing.
• Per-second pricing makes the final cut cheap and the exploration expensive. Track your generation-to-keep ratio; it is your real cost of goods.
• Both reviews attacked story, not image quality. The model has crossed the "looks fine" line, so the bar is now the one every film faces.
• Watch the comments on the Grok Imagine post. A making-of thread from Wonder Studios would be the most useful artifact to come out of this.

Key Takeaways

• Grok Imagine posted Episode 1 of an AI-generated Odyssey on October 23, 2026: 7 minutes 12 seconds, served at 3840x2160, credited to Wonder Studios and director Dev Walker.
• On July 22 Musk promised a "full-length," historically accurate Odyssey movie "before this year ends." The episode arrived 93 days later with no episode count or schedule, leaving 69 days on the promise.
• TechCrunch called the episode "completely airless" and The A.V. Club "a choppy trailer or sizzle reel"; both faulted story and character, not image quality.
• xAI's grok-imagine-video-1.5 generates up to 15 seconds per clip at $0.08 per second; the 432 seconds in the final cut would cost $34.58 at list price if generated once.
• Nolan's The Odyssey stands at $1,794.4 million worldwide on a $250 million budget through October 25, with a 172-minute runtime.
• Wonder Studios, founded in 2025 with more than $15 million raised, is the human production team behind the episode, with a credited director.

Sources: Grok Imagine on X, October 23, 2026, Elon Musk on X, October 24, 2026, Wonder Studios on X, October 23, 2026, Variety, TechCrunch, The A.V. Club, Khaleej Times, July 23, 2026, Far Out Magazine, July 22, 2026, The Numbers: The Odyssey (2026), xAI docs: video generation, xAI docs: models and pricing, Wonder Studios: About