News
Grok's First AI-Made Odyssey Episode Runs Seven Minutes: What It Shows Against a Promised Feature by Year's End
ยท RenderBob team
Grok Imagine posted a seven-minute AI-generated episode of Homer's Odyssey on 23 October. In July, Elon Musk said Grok would make a full-length feature by the end of the year.
Grok Imagine posted a seven-minute AI-generated episode of Homer's Odyssey on X on Friday 23 October. In July, Elon Musk said Grok would make a full-length, "historically accurate" feature by the end of the year.
What was released
According to reports from 24 to 25 October, the episode was posted to Grok Imagine's X account and labelled Episode 1. It shows Odysseus preparing to leave, Penelope urging him to go, the Greek forces sailing for Troy, and cuts to Telemachus growing up and training for combat. The post credits "creative collaboration" with Wonder Studios and a director named Dev Walker. One report says it was made with Grok Imagine Video 1.5, xAI's text-to-video model with synchronised audio. We saw that detail in only one outlet, so treat it as unconfirmed. No outlet we read says how many episodes are planned or when the next arrives.
The promise
Variety's Australian edition reports that on 22 July Musk said Grok Imagine would make a full-length, "historically accurate" version of the Odyssey, "true to the art of Homer". His target was the end of 2026. The only demonstration then was a three-minute AI short made by another X user. The same report questions how "historically accurate" a retelling can be of an epic with monsters and magic.
From 26 October to the end of the year is about nine weeks. One episode of seven minutes does not show whether the other eighty-odd minutes of a feature will follow.
How it was received
Coverage praised the spectacle and criticised the rest. As quoted by LatestLY, TechCrunch's Anthony Ha pointed to a "long voiceover introduction, stilted dialogue and lack of narrative momentum" and said there was "little reason to care about the characters". Mashable noted typical AI-video flaws such as unnaturally perfect teeth and odd skin texture. Other reviewers said the episode lacked human interest and emotional depth, and some described it as reading like a trailer or sizzle reel. These are critics' opinions, and the episode's X viewing figures were reported by only one outlet, so we do not use them.
The release has also been tied by some outlets to Musk's public criticism of another Odyssey film. That dispute is outside what this blog covers.
What a filmmaker can take from it
- Voiceover is a warning sign. A long spoken introduction often stands in for scenes that do not yet carry the story.
- Spectacle scales more easily than story. Large images were praised; dialogue and momentum were not.
- Credit the collaborators. The post names a studio and a director, which is the kind of specific credit our disclosure pieces recommend.
- Judge a promise by its schedule. A first episode is a data point, and the deadline is the test.
What to watch
Whether Episode 2 appears, how long the full piece runs, and whether any of it is described with a full account of tools and human work.
RenderBob 3 and this story
RenderBob 3 lays a long piece out as clips on a timeline, each generated clip on its own lane, so the gap between a seven-minute episode and a feature shows shot by shot, and the model for each clip is your choice, local or cloud.
More from the blog
- YouTube Music Starts Labelling Songs "Created Primarily Using Generative AI": What Is Defined, and What Is Not
YouTube announced on 14 October that YouTube Music will label songs made primarily with generative AI. The threshold, disputes and recommendation effects are unsaid.
- A Report Says Midjourney Reproduced Protected Characters in 37% of Prompts: Why That Number Needs Its Method Before You Use It
Deadline reported that research named Midjourney the worst offender for producing protected Hollywood characters. The 37 percent figure comes from a secondary summary.