Veo 3.1 vs Grok Imagine video: which one can Sume run?

Sume can run Grok Imagine video and cannot run Veo 3.1. The Sume video catalog has no Veo, Sora, Runway, Luma or Midjourney model, so a Veo-versus-Grok test on Sume is really a Grok-versus-other-Sume-models test. This page states what Grok Imagine video accepts here, so you can decide whether it can stand in for the clip you meant to make with Veo.

This post does not compare Veo's quality or price; those are Google's to publish and are not in the Sume docs.

What does Grok Imagine video accept on Sume?

The model id is grok-imagine-video-1.5. It requires an image_url, so it animates a still and does not generate from text alone. Duration is 4 to 15 seconds; resolutions are 480p and 720p; it has no audio, no end frame, and no aspect_ratio or reference inputs. Its list rate is $0.01 per second, and Sume bills list times 1.25, rounded up to the cent.

What if I need audio, a ratio or a text-to-video prompt?

Pick another Sume model. kling-3 takes first and end frames, 16:9, 9:16 and 1:1, and audio on or off. gemini-omni-flash-1.1 always has audio and supports 16:9 and 9:16. seedance-2.5 takes the widest set of ratios and all reference types. Run the same prompt on two of them and compare the clips yourself.

How do I keep the choice swappable?

Keep the model id in config, not in code, and log the id with each job. The create call is the same for every model, so swapping is a one-line change plus a limits check. Read the job record after each run rather than assuming which model served it.