Up to 1080p
Text-to-video
Prompt
Use when no starting frame is required.
Quick answer
Use grok-imagine-video-1.5 for current xAI API video work. Text and single-image generation can reach native 1080p; reference-led generation tops out at 720p. Choose the mode before you design prompts, inputs, storage, and error handling.
xAI; the current API model ID is `grok-imagine-video-1.5`.
Text-to-video, image-to-video, reference-to-video, editing, and extension use separate request shapes.
Standard generation accepts 1–15 seconds. Editing keeps the source duration and caps eligible input at 8.7 seconds.
Text-to-video and image-to-video support 1080p. Reference-to-video and editing are capped at 720p.
xAI's July 31 release documents up to seven reference images per generation.
Jobs are asynchronous. The SDK can poll automatically; REST integrations start a request and poll its request ID.
Not established by xAI's release. Check the current Flowith model selector independently.
Mode decision
Up to 1080p
Prompt
Use when no starting frame is required.
Up to 1080p
Prompt + starting image
Use one image as the first-frame anchor.
Up to 720p
Prompt + reference images or permitted preset voices
Use references for identity or scene consistency, not as a starting frame.
Mode-specific
Prompt + existing video
Editing preserves duration and aspect ratio; extension continues the final frame.
Production gate
The API can return a temporary video URL after asynchronous processing. Your application still owns authorization, retention, review, delivery, and fallback behavior.