Video model comparison
The two lowest-priced tiers in the RelayMe video lineup. One gives you 1080p and fixed clip lengths; the other gives you the lowest per-second rate, a 2K tier and video/audio references.
MiniMax-H3 is the cheaper of the two at 120 credits per second (768P) versus 150 for Grok Imagine, and it is the only one with a 2K tier (200 credits/sec) and reference video and audio inputs. Grok Imagine renders true 1080p at the same 150 credits per second, with fixed 4 / 6 / 8 / 10 / 12 / 15 s clip lengths, seven aspect ratios and the same price with or without audio. Pick MiniMax-H3 for cost and reference-driven shots; pick Grok Imagine for 1080p social clips at a flat price.
| Item | Grok Imagine | MiniMax-H3 |
|---|---|---|
| Credits per second (base tier, with audio) | 150 (480p / 720p / 1080p) | 120 (768P) |
| Credits per second (base tier, no audio) | 150 | — |
| Credits per second (4K / 2K, with audio) | — | 200 (2K) |
| Credits per second (4K / 2K, no audio) | — | — |
| Clip length | 4 / 6 / 8 / 10 / 12 / 15 s | 5–15 s, 1-second steps |
| Resolutions | 480p / 720p / 1080p | 768P / 2K |
| Aspect ratios | 16:9, 9:16, 1:1, 4:3, 3:4, 3:2, 2:3 | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive |
| Audio track | Optional (same price) | Always included |
| Inputs and references | Text-to-video; one first-frame image (up to 1080p) or 2–7 reference images (up to 720p) | Text-to-video; up to 9 reference images, 3 reference videos and 3 audio clips |
| Model family | xAI | MiniMax |
Credit prices come from the same live model list as the pricing page USAGE table; 1,000 credits is about US$1.
Credits for one 8-second clip
| Setting | Grok Imagine | MiniMax-H3 |
|---|---|---|
| Base tier, with audio | 1,200 · ≈US$1.20 | 960 · ≈US$0.96 |
| Base tier, no audio | 1,200 · ≈US$1.20 | Always with audio |
| 4K / 2K, with audio | — | 1,600 · ≈US$1.60 |
For one 8-second clip with audio: Grok Imagine at 1080p is 1,200 credits (about US$1.20), MiniMax-H3 at 768P is 960 credits (about US$0.96), and MiniMax-H3 at 2K is 1,600 credits.
The tables compare settings and costs, not measured visual quality or speed. In one product-image-to-video run on 2026-08-25, the Nano Banana special channel generated a 1:1 product image for 20 credits in about half a minute. Veo 3.1 Fast then generated a 4-second, 720p, 16:9 clip with audio for 600 credits in about a minute. The total was 620 credits. This is one historical observation, not a current quote or a cross-model performance result.
Read the workflow and test context
For your own comparison, keep inputs and prompts fixed and record the model, channel, settings, date, actual credits and elapsed time. Evaluate cost calculations separately from generated results.
Connect one Text node to a Grok Imagine node and a MiniMax-H3 node, attach the same reference image to each, and run once: both clips appear side by side in the task history with their exact credit cost.
Both models bill through RelayMe credits, so there is no separate xAI or MiniMax account to open, and you can add a Veo 3.1 node to the same workflow as a third reference point.
No credit card required · 1,000 credits ≈ US$1
MiniMax-H3, at 120 credits/sec for 768P against 150 for Grok Imagine. An 8-second clip with audio is about US$0.96 versus US$1.20. MiniMax-H3’s 2K tier is 200 credits/sec.
On RelayMe MiniMax-H3 offers 768P and 2K; there is no separate 1080p tier. Grok Imagine offers 480p, 720p, 1080p.
Yes. Grok Imagine offers 4 / 6 / 8 / 10 / 12 / 15 s; MiniMax-H3 offers 5–15 s, 1-second steps. Both top out at 15 seconds, longer than Veo 3.1’s 8.
Yes. Grok Imagine lets you choose whether to generate an audio track at the same 150 credits/sec; MiniMax-H3 always includes a native audio track and it cannot be switched off.
Grok Imagine takes one first-frame image (up to 1080p) or 2–7 reference images (up to 720p). MiniMax-H3 takes up to 9 reference images, 3 reference videos and 3 audio clips per generation, and is the only one of the two that accepts video and audio references.