A pilot landed, and the internet had opinions fast
On October 23, 2026, xAI's Grok Imagine posted the first episode of an AI-generated adaptation of Homer's Odyssey to X. Elon Musk quote-posted it the same night. The pilot runs about seven minutes and 12 seconds, was rendered in 4K at 3840 by 2160, and was built with Wonder Studios and director Dev Walker using Grok Imagine. xAI is framing it as the opening chapter of an ongoing series, not a standalone film.
The project goes back to July 22, 2026, when Musk committed to a historically accurate adaptation before year's end. The pilot quietly walks that back: the dialogue is in English, not Homeric Greek, a detail Variety and other outlets flagged as a concession on the accuracy pitch.
Within 72 hours, TechCrunch, Variety, Mashable, AV Club, and others had weighed in, making this the freshest test case in AI episodic production right now, and a useful one for anyone building a series with similar tools.
The reviews: polished, but not alive
The critical consensus is consistent across outlets. TechCrunch's Anthony Ha called the pilot 'completely airless,' pointing to stilted dialogue and a lack of artistry, inspiration, or human interest. Other reviewers echoed that the episode is technically more polished than prior Musk-posted AI videos, but still suffers from a generic narrator voice and dialogue that doesn't land.
That gap matters for creators: the visual bar is clearly clearable with current tools, but writing, pacing, and performance are still the parts that separate a demo reel from something people want to watch twice. If you're planning your own AI series, assume the hardest problems are not the ones solved yet.
There's also a scale comparison worth sitting with. Christopher Nolan needed six countries and thousands of people to make his version of The Odyssey. xAI is attempting the same material with a video generator, a small production team, and a fraction of the budget. That gap in resources is exactly why the result reads as technically competent but narratively thin — and it's a fair preview of the tradeoffs any solo or small-team creator will hit too.
Engagement was solid, not viral
As read on X around 17:30 UTC on October 25, 2026, Grok Imagine's original post had reached 3,955,704 views, and Musk's quote-post had reached 2,190,746 views. Those are real numbers, but they're modest relative to Musk's typical reach, which suggests curiosity rather than widespread enthusiasm. Nobody involved has said how many episodes are planned or when episode two will arrive, which is itself a signal worth noting if you're tracking the series for benchmarking purposes.
Episode two is the real test, not episode one
A strong pilot proves a method can work once, inside a single production run where the same models, prompts, and reference assets are active the whole time. It does not prove the method can hold up across separate runs, which is the actual job of an episodic series. Episode two is where that gets tested: do the characters' faces stay the same, do voices stay consistent, does the tone of the world hold, or does everything subtly drift once the team starts a new generation session from scratch.
This is a harder problem than it sounds. Holding a character steady for seven minutes inside one continuous pipeline is a different task than reproducing that same character accurately a week or a month later, after different prompts, different settings, or a different take on a scene. Drift in face, voice, or tone across episodes is the single most common failure mode in AI-generated series, and it's usually invisible in a single episode because there's nothing yet to compare it against.
If and when episode two of Odyssey lands, that's the moment to actually watch closely — not for production value, but for continuity. Does the same actor-model render the same character? Does the voice match? Does the setting feel like the same world? That comparison will tell creators more about the real state of AI episodic production than the pilot did.
- Compare character faces side by side between episode one and episode two, not just on first impression
- Listen for voice consistency, not just voice quality
- Check whether sets, lighting, and color tone match or have visibly shifted
- Note any changes in dialogue style or pacing that suggest a different creative pipeline was used
The economics changed, and that's the part creators should act on now
Separate from the creative verdict, there's a practical data point worth taking seriously. Grok Imagine Video 1.5 prices generation at roughly $4.20 a minute, compared to OpenAI's Sora 2 Pro at about $30 a minute, an 86 percent difference. A seven-minute episode at Grok's pricing lands in a completely different budget tier than the same length at Sora's.
One analysis of the release argues this cost curve changes who can realistically attempt a series. If a small team can ship an episode on a regular schedule at this price point, the minimum viable budget for an animated or effects-heavy show drops by orders of magnitude, even though the output isn't studio quality yet. That's a meaningful shift for solo creators and small teams who previously couldn't justify the cost of experimenting with a recurring AI cast.
The caveat is that cheap generation doesn't solve the consistency problem described above. Lower cost per minute makes it more affordable to attempt a multi-episode series, but it doesn't guarantee the character, voice, and tone will hold across those episodes. That's a separate, harder problem, and it's exactly where tools built for maintaining a consistent cast across episodes, like Episodo, become relevant rather than optional.