Hit the line at video speed
Framedash wraps the MiniMax H3 Max text-to-video model for quick drafts. Set resolution, length, and aspect ratio, then watch the lap timer while the provider works.
Generation is paused. You can still draft prompts and preview credit packs.
First-frame and last-frame image-to-video is supported by the model API. Uploads are not open on Framedash yet.
What H3 Max is
Per the public listings, MiniMax H3 Max is a post-trained variant of MiniMax H3, jointly released by MiniMax and fal.ai on 27 August 2026. It targets fast text-to-video and image-to-video (first/last frame) at 480p or 768p, 5–15 seconds at 24 fps.
| Setting | Options |
|---|---|
| Resolution | 480p (832×480) or 768p |
| Duration | 5, 10, or 15 s (whole seconds) |
| Aspect | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 |
| List API price | 480p $0.05/s · 768p $0.08/s |
How it works
- Write a prompt (10–1500 characters).
- Pick 480p or 768p, duration, and aspect ratio.
- We verify Turnstile, reserve daily spend caps, then submit to the API.
- Poll until complete; download only through Framedash (provider URLs stay server-side).
Failed or timed-out jobs (15 min) refund credits or free uses once.
Pricing teaser
Credits map to list provider cost: 480p costs 5 credits per second; 768p costs 8. Packs start at $5.99 for 250 credits. See full pricing.
Compare models: H3 vs H3 Max · H3 Max vs Wan 3.0 · Prompt library
What people ask before they generate
Is Framedash free?
When generation is on, each visitor gets three free renders at 480p and 5 seconds. Higher resolution, longer clips, or extra renders use credits from one-time packs.
How fast is MiniMax H3 Max here?
fal markets the model for speed on its own stack (~3 s for a 5 s clip). Through our OpenRouter path a 480p 5 s clip took about 16 s end to end in our test (October 2026).
Does output include audio?
OpenRouter returns silent video for this model. Plan for your own soundtrack in post.
Who runs the API?
Framedash is independent. We call a third-party API; we are not MiniMax or fal.ai.