MultimodalImage, audio & video
Grok Imagine releases seven-minute first episode of its AI-generated 'Odyssey'
xAI's Grok Imagine posted a seven-minute Episode 1 of The Odyssey, built with Wonder Studios and director Dev Walker, three months after Elon Musk promised a full-length AI feature by year-end. Critics found shots realistic but the storytelling flat.
From the TechCrunch AI article of October 25, 2026 and cited sources: In July, Musk promised a full-length, 'historically accurate' Odyssey made with Grok Imagine before the end of the year. [2] [3] [6]
SOURCES[2] Khaleej Times: After criticising Christopher Nolan, Musk… khaleejtimes.com[3] TeslaNorth: Elon Musk Shares Episode 1 of Grok Imagine's… teslanorth.com[6] Grok Imagine on X: Episode 1 post x.com
AT Pulse research deskPublished · Updated · 5 min read · AI-assisted analysisAI-assisted, not human-edited. Methodology Corrections Who owns this site
- What happened
- In July, Musk promised a full-length, 'historically accurate' Odyssey made with Grok Imagine before the end of the year. [2] [3] [6]
- Original source
- TechCrunch AI article, independent perspective
- Brief by
- AT Pulse research desk AI-assisted; publisher and contact on the desk page
- When
- October 25, 2026 article published
- Covered
- Oct 26 edition
- Evidence
- 13 sources across 10 sites
- 1Grok releases the first seven minutes of its AI-generated…techcrunch.com
- 2Khaleej Times: After criticising Christopher Nolan, Musk…khaleejtimes.com
- 3TeslaNorth: Elon Musk Shares Episode 1 of Grok Imagine's…teslanorth.com
- 4Social Media Today: X offers $175K for Grok-generated…socialmediatoday.com
- 5Basenor: Grok Imagine's Odyssey Contest: 5 Details That…basenor.com
- 6Grok Imagine on X: Episode 1 postx.com
QuietHow it’s measured
Hacker News · 1 like
Reach blends public engagement (views, likes, comments, shares, stars) from the platforms where the story or its original source lives: YouTube, Hacker News, GitHub, Substack and similar. A video and the article written from its transcript share one score, taken from the video. Platforms with no data are left out rather than counted as zero, and the age of a story never lowers its score.
In July, Musk promised a full-length, 'historically accurate' Odyssey made with Grok Imagine before the end of the year. So far the result is one episode of just over seven minutes, posted in 4K and credited to Wonder Studios and director Dev Walker. xAI has not said how many more episodes will follow or when. [2] [3] [6]
TechCrunch found some single shots convincing. It said the episode as a whole lacks momentum, with stilted dialogue, a long generic voiceover and little reason to care about the characters. Reaction on X was split between praise for the polished visuals and complaints that it lacks emotion and coherence. [1] [11]
individual shots are realistic, but a seven-minute episode made under a 15-second-per-clip limit implies dozens of human-edited clips.
Read the full assessment
Implication: long-form AI video still depends heavily on direction, consistency tooling and editing, so budgets should cover orchestration, not just generation fees.
Executive brief
On July 22, Elon Musk promised a full-length, "historically accurate" AI film of The Odyssey by the end of the year. Three months later, the result is a seven-minute "Episode 1" (TechCrunch; Khaleej Times). Grok Imagine made it with Wonder Studios and director Dev Walker. xAI has not said how many episodes will follow or when (TeslaNorth). TechCrunch found single shots convincing but said the piece as a whole had no artistry or momentum. The project is less evidence that AI can make feature films than a public test of how much human direction and stitching long-form AI video still needs.
Read the full section
On July 22, Elon Musk promised a full-length, "historically accurate" AI film of The Odyssey by the end of the year. Three months later, the result is a seven-minute "Episode 1" (TechCrunch; Khaleej Times). Grok Imagine made it with Wonder Studios and director Dev Walker. xAI has not said how many episodes will follow or when (TeslaNorth). The episode drew millions of views. TechCrunch found single shots convincing but said the piece as a whole had no artistry or momentum. The project is less evidence that AI can make feature films than a public test of how much human direction and stitching long-form AI video still needs.
What changed and event timeline
Creator contest
X offered $175,000 in prizes for Odyssey videos made with Grok Imagine. Entries had to run 3–5 minutes with at least one minute of English dialogue. Submissions closed August 31 (;).
Musk promises a feature
He posted that Grok Imagine would make a full-length Odyssey "true to the art of Homer" before the year ended. This came after he attacked the casting of Christopher Nolan's film ().
Episode 1 posted
@imagine published the 4K episode, which runs just over seven minutes. It credited Wonder Studios and Dev Walker and offered to post a breakdown of how it was made (;).
Critical review
TechCrunch called the clip "airless," with stilted dialogue and a generic narration, though some individual shots were realistic ().
Reach numbers
Cryptopolitan reported more than 6 million views, 19,000 likes and 2,100 reposts ().
Capabilities and access
- Model: The release post says only "made with Grok Imagine" (X post).
- xAI's published model:
grok-imagine-video-1.5-preview, announced June 3, 2026. - Per-generation limits: up to 15 seconds per generation. Output costs about $0.08 per second (PixVerse summary of xAI docs).
Read the full section
- Model: The release post says only "made with Grok Imagine" (X post). Secondary outlets name Grok Imagine Video 1.5 (Basenor). xAI has not confirmed which model version was used.
- xAI's published model:
grok-imagine-video-1.5-preview, announced June 3, 2026. It turns a still image into video, outputs up to 720p and is available through the xAI API (xAI). - Per-generation limits: up to 15 seconds per generation. Output costs about $0.08 per second (PixVerse summary of xAI docs).
- Consumer access: SuperGrok subscribers get daily credits, and free X users get a limited number (Basenor).
Technical analysis for researchers and developers
- Stitched shots, not one generation. That is an inference from the documented limits.
- The episode was posted in 4K, but xAI documents only 720p output for the preview model.
- In the week of the release, Grok Imagine published a step-by-step guide to creating characters (TeslaNorth).
Read the full section
- Stitched shots, not one generation. At 15 seconds or less per generation, a seven-minute episode needs roughly 30 or more separate generations, edited together by people. That is an inference from the documented limits. The contest rules point the same way: they allowed outside tools for editing, music and sound effects (Social Media Today).
- Resolution gap. The episode was posted in 4K, but xAI documents only 720p output for the preview model. This suggests upscaling or another pipeline, which is also an inference.
- Character consistency. In the week of the release, Grok Imagine published a step-by-step guide to creating characters (TeslaNorth). That points to reference-image workflows for keeping characters consistent across shots.
- Reproducibility. There is no published prompt log, shot list or count of human edits. The promised breakdown has not appeared.
Claims and evidence
- "Historically accurate": This is Musk's claim and it has not been substantiated. Critics note that the epic includes a cyclops and a witch (Khaleej Times).
- Video 1.5 ranks first in blind Arena text-to-video tests, ahead of Sora 2, Veo 3.1 and Kling: This is vendor-reported, and independent verification is still pending (Basenor).
- Engagement: The view counts come from X metrics as reported by one outlet (Cryptopolitan). They measure reach, not quality.
Read the full section
- "Historically accurate": This is Musk's claim and it has not been substantiated. Critics note that the epic includes a cyclops and a witch (Khaleej Times). TechCrunch points out that none of the characters speak ancient Greek (TechCrunch).
- Video 1.5 ranks first in blind Arena text-to-video tests, ahead of Sora 2, Veo 3.1 and Kling: This is vendor-reported, and independent verification is still pending (Basenor). It also conflicts with documentation saying the preview model does not support text-to-video (PixVerse).
- Engagement: The view counts come from X metrics as reported by one outlet (Cryptopolitan). They measure reach, not quality.
Context and prior work
- The Nolan film: Released in July, it was mostly shot practically across six countries.
- The casting dispute: Musk objected to Lupita Nyong'o being cast as Helen of Troy (Khaleej Times).
- Contest winner: Nem Perez won the $100,000 first prize with a five-minute entry about the Phaeacians and Nausicaa (Basenor).
Read the full section
- The Nolan film: Released in July, it was mostly shot practically across six countries. It sold out Imax screenings for months and is a likely Oscar front-runner (TechCrunch).
- The casting dispute: Musk objected to Lupita Nyong'o being cast as Helen of Troy (Khaleej Times).
- Contest winner: Nem Perez won the $100,000 first prize with a five-minute entry about the Phaeacians and Nausicaa (Basenor).
Limitations, safety and contested findings
- Craft: TechCrunch says the narrative is weak: a long voiceover, one bearded face after another and no reason to care about the characters (TechCrunch).
- Reception is split: Some viewers praised the polished visuals. Others said it lacks emotion and coherence (X trending summary).
- Pricing conflict: Social Media Today has a version of its contest story headlined "up to $200K" (Social Media Today).
Read the full section
- Craft: TechCrunch says the narrative is weak: a long voiceover, one bearded face after another and no reason to care about the characters (TechCrunch).
- Reception is split: Some viewers praised the polished visuals. Others said it lacks emotion and coherence (X trending summary).
- Pricing conflict: Social Media Today has a version of its contest story headlined "up to $200K" (Social Media Today). This is not reconciled with the $175,000 figure. A secondary figure of about $4.20 per minute also differs from the $0.08-per-second rate (Digital Applied).
Business and practitioner implications
- Short clips are viable, long stories are not yet. Realistic shots are achievable now.
- Teams should plan for shot planning, consistency tooling, upscaling and editing, not just per-second generation fees.
- The deadline is tight.
Read the full section
- Short clips are viable, long stories are not yet. Realistic shots are achievable now. Sustaining a narrative still depends on human directors, here Wonder Studios and Dev Walker.
- Budget for orchestration. Teams should plan for shot planning, consistency tooling, upscaling and editing, not just per-second generation fees.
- The deadline is tight. A feature by December 31 would mean producing more than ten times the footage released so far, at a quality critics have not yet accepted.
- Validate before choosing a vendor. Run your own head-to-head tests rather than relying on xAI's Arena claims.
The source trail.
Sources (13)
Grok releases the first seven minutes of its AI-generated ‘Odyssey’
Article text retrieved; extracted text may omit tables or interactive elements.
techcrunch.com