Grok publishes first seven minutes of its AI-generated Odyssey

Grok has published seven minutes of AI-generated footage based on Homer's "The Odyssey". The company calls it "Episode 1 of our Odyssey series." The clip was posted on X on October 25, 2026, with the note that it was "made with Grok Imagine" in collaboration with Wonder Studios and director Dev Walker.

The release falls short of the original promise. On July 22, Elon Musk wrote on X that "before this year ends, Grok Imagine will make a full-length movie of The Odyssey," one that would be "historically accurate and true to the art of Homer," according to Khaleej Times. About two months of that deadline remain. What exists so far is one episode, and the project is now described as a series rather than a film.

What the footage shows

TechCrunch weekend editor Anthony Ha reviewed the clip. He writes that it is "not as spectacularly ugly" as he had expected. At moments, he says, he might have believed he was watching real actors on a real sea. On a technical level, that is the notable part. Seven minutes of generated footage holds together well enough that some individual shots pass as live action.

The weaknesses are in the parts that make footage a story. Ha describes a long narrated backstory, delivered in a generic British accent, followed by "one bearded face after another, each shouting stilted lines." He calls the whole "completely airless," with "no sense of momentum, no reason to care about anyone or anything on-screen." He says he was checking the remaining runtime before the halfway point, and that he would have skipped it if it were a video game cutscene.

The "historically accurate" claim is also hard to test. No one in the clip speaks ancient Greek, Ha notes. The source is a poem about gods, monsters and a witch. That makes historical accuracy a loose standard to begin with.

The contrast with Nolan's film

The project began as a reaction. Musk had criticised Christopher Nolan's "The Odyssey" over its casting. Khaleej Times reports that Musk agreed with a post objecting to Lupita Nyong'o being cast as Helen of Troy, and had earlier called the casting "an insult."

The two productions are very different ways of making a film. According to TechCrunch, Nolan's version used plenty of visual effects but was mainly practical filmmaking. It was shot in six countries with a cast and crew of thousands. It was released in July, sold out Imax screenings for months, and is likely to be the front-runner at the next Academy Awards. Nolan has called himself "a techno-skeptic" and has said that before AI he had never seen a technology "so completely rejected by the public."

The Grok version credits one studio and one director, working with a generation model. Neither the post nor the TechCrunch review says how many people worked on it, how long it took or what it cost. The post asks viewers whether they would like "a full breakdown of how it was done," which suggests the method has not been published yet.

What seven minutes measures

Long-form AI video has mostly been judged by its technical failures: faces that drift, physics that break, shots that do not match from one cut to the next. This clip suggests those problems are getting smaller over several minutes of continuous narrative. Ha's main complaints are not about glitches. They are about pacing, dialogue, voice and character.

Those are writing and directing problems. In a conventional production they belong to screenwriters, editors, actors and directors. The clip shows that a model can now produce something that looks like an epic. It does not show that a model can supply the judgement that makes an audience keep watching. It also leaves open how much of the result came from Dev Walker and Wonder Studios rather than from Grok Imagine.

What it means for work

The tasks this footage takes over are in physical production: casting extras, building sets, filming at sea, lighting and photographing faces. For background plates, previsualisation and short inserts, generated shots at this level could plausibly replace some location and crew work. The tasks it does not take over are the ones the review faults: structure, dialogue, performance and pacing. Those still depend on people, and here they were thin. A human director was credited, so the clip is evidence of a new production method more than of filmmaking without filmmakers. What remains open is cost, labour hours and how much manual correction went into each shot. Those numbers would show what this method actually changes for crews.