SpaceXAI prices Grok Imagine Video 1.5 Lite by the second

SpaceXAI lists a lower-priced video model, grok-imagine-video-1.5-lite, in its developer docs. It showed up in public posts this week. Vercel said on 8 October 2026 that it was live on AI Gateway as spacexai/grok-imagine-video-1.5-lite. Artificial Analysis posted a ranking the same evening, at 22:23 UTC.

Text, a still, and what stays on the full model

Text-to-video means the prompt describes the shot, and the model makes a first frame from that description, then animates it. Image-to-video means a still you pass in becomes the first frame, and the prompt describes what happens next. SpaceXAI’s video overview says Lite does both, with lip-synced speech, in clips of 1 to 15 seconds, at 480p, 720p, or 1080p. It says 1080p is upscaled from 720p. Full grok-imagine-video-1.5 renders 1080p natively, and it keeps the controls Lite leaves out: up to 14 reference images, up to three voice references, and pinned first, last, and mid-video frames. The overview points developers to Lite for simple videos.

Prices, and a 10-second example

The Lite model page prices output per second by resolution: $0.02 at 480p, $0.03 at 720p, and $0.14 at 1080p. Image input is listed at $0.01. The text-input row has no dollar amount. The full 1.5 page uses the same layout at $0.08, $0.14, and $0.25 for 480p, 720p, and 1080p, with image input again at $0.01. The pricing index prints one number per model, $0.020 per second for Lite and $0.080 per second for full 1.5, which matches the 480p row only. Use the resolution table for a quote.

A 10-second clip at 720p is 10 times $0.03, or $0.30, before any image. The same length at 1080p is 10 times $0.14, or $1.40. An image-to-video call adds the $0.01 image input. Those are the printed rates.

Artificial Analysis, and Google’s Veo rates

Artificial Analysis says Lite is number 17 on both AA-Video-T2V v2.0 and AA-Video-T2V-Silent v2.0, “ahead of Google’s Veo 3.1 at about a third of its price.” A Pareto frontier, on a chart of quality against speed, is the set of results where nothing else is both better and faster. Artificial Analysis says Lite is the fastest model at its quality level, which is why it puts Lite on that frontier, at a median of 60.5 seconds for a 10-second 1080p clip. The rank and the timing are Artificial Analysis’s own method. Its text-to-video leaderboard lists the model among recent additions. Different views on that page rank the same model differently, so the 17 here is the figure Artificial Analysis stated for those two boards. It also says the model is on the SpaceXAI API, on fal at the same prices, and on Vercel AI Gateway. The fal price is Artificial Analysis’s statement.

Google’s pricing page lists Veo 3.1 as video with audio, paid tier, per second. Standard is $0.40 at 720p and 1080p, and $0.60 at 4k. Fast is $0.10 at 720p, $0.12 at 1080p, and $0.30 at 4k. Lite is $0.05 at 720p and $0.08 at 1080p, with 4k not supported. $0.14 is about a third of the $0.40 standard rate, and higher than Google’s Fast and Lite 1080p rates. SpaceXAI’s 1080p is upscaled from 720p. Google’s page does not say whether its 1080p is upscaled. The third-of-the-price line matches the standard with-audio rate, and not every Veo 3.1 price on the page.