Grok Imagine Video 1.5 vs Veo 3.1 (Quality)

Specs side by side

Price per unit and per clip

Prices exclude taxes and are charged from your prepaid USD balance. Failed generations cost nothing.

Grok Imagine Video 1.5 animates a single photo into a 6-15 s clip with native audio, at 480p or 720p, priced per second. Veo 3.1 (Quality) always makes an 8-second clip, from text or from images, at 720p, 1080p or 4K, priced per clip. When I have a photo and want the character to speak, I pick Grok: 8 seconds at 720p come out about 1.8 times cheaper than a Veo clip. Without a photo, or when you need 1080p or 4K, Veo is the option.

Key differences

• Input. Grok Imagine Video 1.5 is image-to-video only: it needs exactly one image in image_url. Veo 3.1 (Quality) generates from text alone or from 1-3 images in image_urls used as frames.
• Length. Grok makes 6-15 s in one-second steps. On Jantaris, Veo always returns 8 s and bills it as one clip (duration is fixed).
• Resolution. Grok has 480p and 720p. Veo has 720p, 1080p and 4K.
• Audio. Grok generates a synchronized track with dialogue, and generate_audio: false does not lower the price. The Veo form has no audio toggle and no audio surcharge. At the vendor, Veo 3.1 generates sound together with the picture, but listen to one clip before you run a batch.
• Frames and ratios. Grok: 16:9, 9:16 and 1:1. Veo: 16:9 and 9:16 only. Prompt: 10,000 characters in Grok, 4,000 in Veo.
• Price for 5 and 10 s. Neither makes a 5 s clip: Grok starts at 6 s and Veo is fixed at 8 s. Grok's shortest clip (6 s) costs about 4.8 times less than a Veo clip at 480p and about 2.4 times less at 720p. 10 s of Grok at 720p is about 70% of one Veo clip; only 15 s at 720p comes out slightly above a Veo clip.

When to pick Grok Imagine Video 1.5

• A talking portrait or an animated cover from an existing photo.
• A spoken line with audio already in the file.
• A shot longer than 8 s, up to 15 s, from a photo.
• Square 1:1 for the feed.

When to pick Veo 3.1

• A clip from text alone, with no starting photo.
• A 1080p or 4K shot for an ad or a trailer.
• A transition across several images given as frames.
• A sequence cut from equal 8-second pieces.

Cost in practice

Grok is priced per second (from $0.088 per second at 480p; 720p costs exactly twice that), Veo per 8 s clip (from $2.53 per video at 720p and 1080p). For 8 seconds at 720p Veo is about 1.8 times more expensive, and against Grok at 480p the gap grows to about 3.6 times. Veo in 4K costs 1.5 times its 1080p price, but Grok has no 4K at all. Price any length in the AI cost calculator; all rates are on /pricing, with worked examples in Grok Imagine API and Veo 3.1 API pricing.

In my September 28 test Grok Imagine Video 1.5 came first with an average of 8.5/10 (coffee 8.0, a hard Polish sentence 9.0), and Veo 3.1 scored 8.0 (coffee 8.0, speech 8.0). I tested both through other providers' APIs, and the Veo form on Jantaris has no audio toggle, so I do not carry the speech result over one to one. Full write-up in the best AI video generator in 2026.

FAQ

Grok Imagine Video 1.5 or Veo 3.1: which is better for dialogue?

In my test Grok did better: both speech recognition models read the hard Polish sentence without an error. The condition is a photo of the character, because Grok does not generate from text alone.

Can Grok Imagine Video 1.5 make a clip from text only?

No, it needs one image in image_url. For text-only clips use Veo 3.1 or Grok Imagine 1.0 Video.

How much do 10 seconds cost in each model?

In Grok, 10 times the per-second rate at the chosen resolution. In Veo, 10 seconds means two 8 s clips, since there is no other length. You see the exact amount before you start, in Studio and in estimated_cost, and a failed generation costs $0.

Vendor examples

Official examples from the model vendors, not our generations.