Christopher Nolan’s “The Odyssey” opened in July and kept selling out Imax screens for months. It was shot across six countries with a cast and crew in the thousands, and it has been the front-runner for the next Academy Awards since the week it landed. Elon Musk did not care for the casting.
In July he said Grok would answer with a full-length, historically accurate adaptation of Homer before the year was out. On Oct. 23, the Grok Imagine account on X posted seven minutes of it and called the clip “Episode 1 of our Odyssey series.” The credits name Wonder Studios and the director Dev Walker. It has drawn roughly four million views in two days.
What the seven minutes contain is a great deal of coastline, a good deal of voiceover and a cast of bearded men shouting at each other in English. The narrator speaks in the rounded British accent that has scored fantasy video games and Hollywood epics for a quarter century. Nobody speaks ancient Greek. Staging is largely absent: figures stand in the frame and deliver lines, and the camera moves because the model was asked to move it.
Viewers were unkind about the acting. Comment threads under the post settled quickly on the word stiff, and on the sense that nothing in the frame was happening to anyone. A handful of individual shots hold up — a hull against a grey sea, a fire on a beach at dusk — but the sequence has no momentum. It is a slideshow with a soundtrack, and the soundtrack explains what the pictures have not established.
The comparison with Nolan is the one Musk invited, and it is not flattering in the way he seems to intend. Nolan has described himself as a techno-skeptic and has said that before AI he had never seen a technology so completely rejected by the public. His film spent real money on real water. Musk’s spent compute.
The timing is its own argument. Musk made the promise in July, when Nolan’s film was setting the cultural agenda, and the seven minutes landed three months later, after the theatrical run had crested. A feature would have taken years at any studio, and he never described a production schedule. What he described was a competitor’s failure to cast the way he would have cast.
Actors’ unions have spent that same three years negotiating over exactly this product. The agreements that ended the 2023 strikes require consent and payment for digital replicas of performers, and they say nothing about a film with no performers at all. A clip like this one sits outside those contracts, which is a labour question rather than a legal one, and one that no court has been asked to settle.
What the release actually demonstrates is the state of video generation in late 2026: shot-level quality is high enough to pass at a glance, and sequence-level craft is missing. Models have learned what things look like. They have not learned why a camera should move, when a cut should land, or how a performance builds across ninety seconds. Those are the parts of filmmaking that live in the edit, and an edit is a decision about what to leave out.
The business logic of the episode is easy to read. A seven-minute clip released free on X costs a fraction of a feature and generates argument, which is the currency Musk trades in. Calling it Episode 1 also buys time: a series has no obligation to conclude, and the promise of a finished film can be renewed indefinitely without ever being tested.
Hollywood has been arguing about this for three years, and the argument has moved from whether the tools work to what they are for. Studios use generated video for storyboards, previsualisation and pickup shots, where a cheap approximation of a scene saves a day of shooting. The gap between that and a feature is not resolution. It is the work of hundreds of specialists whose judgement is the product.
There is also the question of what these systems were trained on, which no major video model has answered in detail. Musk’s public position has been that existing copyright rules are too generous to studios. That position is convenient for a company that would benefit from a looser reading, and it makes the output harder to license for anyone who wants to sell tickets.
For Homer, the damage is limited. The poem has survived worse translations, and its episodes were always told aloud by someone with a voice and a hall to fill. Seven minutes of stilted dialogue on a phone screen is a small addition to that record.
For Musk, the release is a smaller achievement than the announcement implied, and the gap is now measurable. He promised a film and delivered a trailer for a series that has not been commissioned. The interesting question is not whether Grok can generate another six episodes. It is whether anyone involved can tell a story that holds for ninety minutes, which is a skill no one has yet shown a model how to acquire.
The film Musk was reacting to is still in theatres. His answer runs seven minutes and is free to watch. Those two facts say more about the contest than any of the reviews did.
