What is Grok Imagine Video 1.5?
Grok Imagine Video 1.5 is the video generation model from SpaceXSI, the company formerly called xAI. It went into preview at the end of May 2026 and became generally available on 16 June 2026 as grok-imagine-video-1.5 on the API, on grok.com/imagine and in the Grok apps. It makes video with synchronised audio, including sound effects, ambience and lip-synced speech, in the same pass.
It does text-to-video, image-to-video and reference-to-video. A 31 July update added native 1080p for text-to-video and image-to-video; reference-to-video stays at 720p. Clips run from 1 to 15 seconds, with 8 as the default, in nine aspect ratios. Version 1.5 is the only Imagine model that can pin the first and last frames, set up to four keyframes and take voice references for up to three speakers.
According to Tech Times, API pricing is $0.08 a second at 480p, $0.14 at 720p and $0.25 at 1080p. A Lite version from $0.02 a second reaches 1080p by upscaling 720p.
How much does Grok Imagine Video 1.5 cost?
| Plan | Price | What you get |
|---|---|---|
| 480p | $0.08 per second | |
| 720p | $0.14 per second | |
| 1080p | $0.25 per second | |
| Lite | From $0.02 per second |
API, per second of video: $0.08 at 480p, $0.14 at 720p, $0.25 at 1080p; grok-imagine-video-1.5-lite from $0.02. In the Grok apps through SuperGrok and X subscription plans.
Is Grok Imagine Video 1.5 worth it?
Grok Imagine Video 1.5 covers more of a short-video job in one call than most rivals: picture, motion, dialogue and sound arrive together, and image-to-video, reference-to-video, editing and extension sit on one API. Clips run up to 15 seconds. Since the 31 July update, text-to-video and image-to-video render natively at 1080p, up from a 720p ceiling at launch.
Control is the strongest part of the spec. On 1.5 a developer can pin the first and last frames, place up to four keyframes on a one-third-second grid, and give voice references for up to three speakers; Tech Times reports up to seven reference images at once. SpaceXSI says the Fast variant makes a 6-second 720p clip in about 25 seconds, against more than 40 for the previous model. SpaceXSI has published no benchmark scores of its own.
Price depends sharply on resolution. At $0.08 a second for 480p, $0.14 for 720p and $0.25 for 1080p, according to Tech Times, an hour of finished 1080p footage costs $900 before retakes. The Lite model at $0.02 a second is the cheaper route to drafts, but its 1080p is upscaled 720p. Generation is asynchronous, results come back as temporary URLs that must be downloaded promptly, and editing stays on the original model, capped at 8.7-second sources and 720p.
The maker's record is the larger question. In January 2026 Grok's image tools were used to undress real people from uploaded photos, including children, which led to investigations by the European Commission and India, and Malaysia briefly blocked Grok. In April NBC News found Grok still producing sexualised images and videos of real women, including by turning real photos into videos, and reported that eight agencies, among them California's attorney general, the European Commission and the UK's Ofcom, were continuing their investigations. Voice references now let a user keep a real person's face and voice across scenes; Tech Times notes the announcement said nothing about consent safeguards, and API access to voice references goes through SpaceXSI's sales team. The API follows SpaceXSI's general terms: no training on inputs or outputs without permission and 30-day retention for audits, and the US-only endpoint does not serve video.
Where it earns its place
A two-person studio making product teasers for e-commerce brands
Turn a product photo into a 10-second clip with sound
Image-to-video on 1.5 uses the photo as the first frame and generates ambience and voice-over together; a 10-second 720p take costs $1.40, so drafting on Lite first keeps retakes cheap.
An indie game developer storyboarding cut-scenes
Keep two characters' faces and voices consistent across shots
Reference-to-video with character images and voice references keeps them stable, but it tops out at 720p and API voice references need SpaceXSI's sales team.
A trust-and-safety lead at a marketing agency
Decide whether a client campaign can use the model
Weigh SpaceXSI's open investigations and the lack of published consent checks for voice references; real people's likenesses need written releases whatever the tool.
Before you commit
- Setup
- Create a key at console.x.ai and post to the video generation endpoint with grok-imagine-video-1.5; poll the request_id until the clip is ready, which the SDK waits up to 10 minutes for.
- Integrations
- xAI API with xai-sdk and the Vercel AI SDK; separate edit and extension endpoints; also listed on Vercel's AI Gateway.
- Data and privacy
- SpaceXSI API terms: no training on inputs or outputs without permission, 30-day retention for audits, optional team-wide zero retention that disables stored media; outputs are temporary URLs.
- Cost at scale
- 60 minutes of finished video a month costs about $900 at 1080p or $504 at 720p, per Tech Times, before failed takes.
- Limits
- 1 to 15 seconds per clip; reference-to-video at 720p; edits keep the source length up to 8.7 seconds; no video on the US-only endpoint.
What’s good
- Video with synchronised sound effects and lip-synced speech
- Native 1080p for text-to-video and image-to-video
- First and last frames, four keyframes and three voice references
- Fast variant renders 6 seconds of 720p in about 25 seconds
- Lite model from $0.02 a second for drafts
What’s not
- $0.25 a second at 1080p, $900 an hour of footage
- Maker under deepfake investigations on several continents
- No published consent safeguards for voice references
- Editing capped at 720p and 8.7-second sources
- No benchmark report from SpaceXSI; no US-only video endpoint
Best for
Developers and studios who want short clips with sound and fine control over frames, and whose brand can carry SpaceXSI's record.
Grok Imagine Video 1.5 alternatives
Grok Imagine Video 1.5: common questions
- Is Grok Imagine Video 1.5 worth it?
- Capable, fast and reasonably priced video with sound, from a company whose image tools regulators on several continents are still investigating. We rate it 3.6 out of 5.
- How much does Grok Imagine Video 1.5 cost?
- API, per second of video: $0.08 at 480p, $0.14 at 720p, $0.25 at 1080p; grok-imagine-video-1.5-lite from $0.02. In the Grok apps through SuperGrok and X subscription plans.
- What are the downsides of Grok Imagine Video 1.5?
- $0.25 a second at 1080p, $900 an hour of footage; maker under deepfake investigations on several continents; no published consent safeguards for voice references; editing capped at 720p and 8.7-second sources; no benchmark report from SpaceXSI; no US-only video endpoint.
- Who is Grok Imagine Video 1.5 best for?
- Developers and studios who want short clips with sound and fine control over frames, and whose brand can carry SpaceXSI's record.
- How much does Grok Imagine Video 1.5 cost for a team or in production?
- 60 minutes of finished video a month costs about $900 at 1080p or $504 at 720p, per Tech Times, before failed takes.
- Is Grok Imagine Video 1.5 safe for company data?
- SpaceXSI API terms: no training on inputs or outputs without permission, 30-day retention for audits, optional team-wide zero retention that disables stored media; outputs are temporary URLs.
- What are the best Grok Imagine Video 1.5 alternatives?
- The closest alternatives we have reviewed are Veo 3.1, Kling 4.0, Runway Gen-4.5 and Seedance 2.5.
Answers checked Nov 6, 2026.