Grok Imagine Image 2.0xAI · Image · per image
xAI's image model for text-to-image and edits from up to five pictures
- Up to 5 reference images
$0.028
Try3 models · from 18 credits
Grok Imagine Image 2.0xAI · Image · per image
xAI's image model for text-to-image and edits from up to five pictures
$0.028
TryGrok Imagine Video 1.5xAI · Video · 480p · 5s
1 to 15-second clips with sound and spoken lines, timed keyframes and seamless loops
$0.14
TryThe xAI group holds Grok Imagine Image 2.0, Grok Imagine Video 1.5, Grok TTS and Grok STT. Together they cover pictures, short videos with sound, reading text aloud and turning speech back into text. Grok Imagine Image 2.0 makes images from plain sentences and edits or combines up to five pictures. Grok Imagine Video 1.5 makes clips of 1 to 15 seconds with a matching soundtrack and spoken lines. Grok TTS reads text in 71 voices across 20 languages, and Grok STT writes transcripts with word timings in 25 languages. Because they share one maker, they work well as a chain: draw a frame, animate it, add a voiceover, and transcribe the result for captions.
For images, describe the subject, setting, light and style in plain sentences, up to 8,000 characters, and choose one of five frame shapes from square to 16:9 or 9:16. To edit, upload one to five pictures and refer to them as Image 1, Image 2 and so on, saying what to change and what to keep. For video, write the scene, the motion and any spoken line in quotes. Clips run 1 to 15 seconds, six by default, at 480p, 720p or 1080p, and sound is on by default; a silent clip costs the same. Starting from a picture adds the most control: set a last frame, add up to four in-between frames at the seconds you choose, or turn on loop so the clip ends where it began. A third mode builds a scene from one to seven reference images.
Grok TTS reads up to 15,000 characters at a time. Speech tags such as [laugh] or a whispered passage add delivery cues, the speed runs from 0.7 to 1.5, and the language can be set or detected from the text. MP3 is the default, with WAV and raw formats for phone systems. Grok STT accepts recordings from one second to two hours and returns a transcript with word timings, which makes captions and searchable notes easy. List up to 100 names and terms so they are spelled your way, keep or drop filler words, and turn on multichannel when each speaker sits on a separate channel. If numbers, dates and prices should come out as digits, set the language and turn on text normalization.
Clips run from 1 to 15 seconds in whole seconds, with six as the default. Text and picture starts offer 480p, 720p or 1080p, and higher tiers cost more per second. The reference mode, which builds a scene from one to seven images, offers 480p or 720p.
Yes. Start from a picture and turn on loop, and the clip ends on the same frame it started with, so it can repeat smoothly. Loop cannot be combined with a last frame. You can still add up to four in-between frames to guide what happens in the middle.
Yes, by default. Grok Imagine Video generates a matching soundtrack, and spoken lines you put in quotes in the prompt are voiced in the clip. Turning sound off returns a silent video at the same price, which is useful when you plan to add your own music or narration.
Grok TTS offers 71 voices across 20 languages, and it can detect the language from the text. Add speech tags such as [laugh] for delivery, adjust speed from 0.7 to 1.5, and choose MP3, WAV or raw formats for phone systems. Each request reads up to 15,000 characters.
It transcribes audio from one second to two hours in 25 languages and returns word timings for captions. Add up to 100 names and terms so they are spelled correctly, choose whether to keep filler words, and use multichannel when each speaker was recorded on a separate channel.