Nano Banana
MiniMax H3 Max — #1 Image-to-Video Model on Artificial Analysis
MiniMax H3 Max is a post-trained version of the open-weight MiniMax H3, built by fal with new training data and a co-designed inference stack. It debuted at #1 in Image to Video and #3 in Text to Video on the Artificial Analysis leaderboards in August 2026.
- MiniMax H3 Max
- #1 Image to Video on Artificial Analysis
- Native audio
- Up to 9 reference images
More models
- Veo 3.1GoogleText to video · Frames · Up to 4K · Up to 8s
- Veo 3.1 FastGoogleText to video · Frames · Up to 4K · Up to 8s
- Seedance 2.5ByteDanceText to video · Frames · References · Up to 720p · Up to 30s
- Seedance 2.0ByteDanceText to video · Frames · References · Up to 1080p · Up to 15s
- Seedance 2.0 FastByteDanceText to video · Frames · References · Up to 720p · Up to 15s
- Seedance 2.0 MiniByteDanceText to video · Frames · References · Up to 720p · Up to 15s
- Seedance 1.5 ProByteDanceText to video · Frames · Up to 1080p · Up to 12s
- Kling 3.0Kling AIText to video · Frames · References · Up to 1080p · Up to 30s
- Kling 2.6Kling AIText to video · References · Up to 1080p · Up to 30s
- Wan 3.0AlibabaText to video · Frames · References · Up to 1080p · Up to 15s
- Wan 2.7AlibabaText to video · References · Up to 1080p · Up to 15s
- Wan 2.6AlibabaText to video · Frames · References · Up to 1080p · Up to 15s
- MiniMax H3 MaxMiniMaxText to video · Frames · References · Up to 768P · Up to 15s
- MiniMax H3MiniMaxText to video · Frames · References · Up to 2K · Up to 15s
- Hailuo 2.3MiniMaxText to video · Frames · Up to 1080p · Up to 10s
- Hailuo 2.3 FastMiniMaxFrames · Up to 1080p · Up to 10s
- Gemini Omni 1.1 FlashGoogleText to video · Frames · References · Up to 4K · Up to 10s
- Grok Imagine VideoxAIText to video · Frames · Up to 720p · Up to 30s
- HappyHorse 1.1AlibabaText to video · Frames · Up to 1080p · Up to 10s
Why MiniMax H3 Max Is a Great AI Video Generator
H3 Max takes the H3 base and tunes it for prompt adherence and real-world quality. If image-to-video is your main workflow, it is one of the strongest options here.
Leaderboard #1
#1 in Artificial Analysis Image to Video at launch, #3 in Text to Video.
Better prompt adherence
Post-training focused on following prompts more closely.
Audio included
Every clip comes with generated sound.
Reference images
Guide the clip with up to 9 images.
What MiniMax H3 Max can do
Image to video
Use a first frame, or a first and last frame.
Text to video
Describe a scene with sound.
Reference to video
Steer characters and products with up to 9 images.
5–15-second clips
Pick a length from 5 to 15 seconds, with sound generated in the same pass.
480p and 768p output
Choose 480p and 768p in 6 aspect ratios: 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16.
Reference inputs
Besides a text prompt you can add a first frame and an optional last frame and up to 9 reference files.
Use cases
Starting points you can copy into the generator above.
| Use case | Good for | Example prompt |
|---|---|---|
| Animate key visuals | Turn a hero image into a shot. | The astronaut in the image turns to look at the camera as Earth rises behind, visor reflecting the light, low hum |
| Product films | Product photos into motion. | The watch in the image ticks as the camera orbits, water droplets on the glass, soft ambient tone |
| Story beats | From a start frame to an end frame. | Transition from the seed in the first frame to the blooming flower in the last frame, time-lapse feel |
| Character scenes | Characters from references. | The two characters in the reference images share an umbrella in a rainy Tokyo street at night, neon reflections, rain sounds |
MiniMax H3 Max in 3 steps
- 1
Describe the shot or add a frame
Write what happens, how the camera moves and how it should sound. Add a start image or references to lock the look.
- 2
Pick length, resolution and ratio
Set duration, resolution and aspect ratio. The credit cost is shown before you submit.
- 3
Generate and keep your results
Videos appear in the generator and are saved to My works. Anything that fails is refunded automatically.
X Posts About MiniMax H3 Max — #1 Image-to-Video Model on Artificial Analysis
- Artificial Analysis@ArtificialAnlys
MiniMax H3 Max, a post-trained version of MiniMax H3 developed by fal, debuts at #1 in Image to Video and #3 in Text to Video on the Artificial Analysis Video Leaderboards with Audio, ahead of the base MiniMax H3 on both MiniMax H3 Max is built and served by fal, and… pic.twitter.com/QGa01D9GLF
View on X - MiniMax (official)@MiniMax_AI
Some milestones move a leaderboard. A rare few change how you see the future. MiniMax H3 Max did both. We released MiniMax H3 as an open-weight model with the hope that the community would adapt it, challenge it, and take it further than we could alone. @fal did exactly that.…
View on X - MiniMax (official)@MiniMax_AI
Congrats to the fal team on MinMax-H3 Max - fantastic work! fal optimized the model for stronger real-world performance while co-designing the inference system around it to make sure faster than real-time is still possible. This required two capabilities that rarely sit under the… https://t.co/PM4b2yyUVY
View on X - MiniMax (official)@MiniMax_AI
A great deep dive from the @fal team on how they built H3 Max, combining post-training with a co-designed inference stack to improve prompt adherence, visual quality, and speed. We're proud to see MiniMax H3 serve as the foundation for this work. Congratulations to the team, and… https://t.co/4ON1xReRYZ
View on X - AGTP@AGTPinsights
MiniMax H3 Max just let a developer stream infinite AI-generated Rick and Morty-style TV today. Here's what you need to know. Rehan Sheikh hooked https://t.co/vNMENfacw9's new MiniMax H3 Max model to a Twitch livestream, calling it infinite interdimensional cable. The model… pic.twitter.com/qRwRnjwhp1
View on X
YouTube Videos About MiniMax H3 Max — #1 Image-to-Video Model on Artificial Analysis
MINIMAX H3 JUST GOT EVEN BETTER!
Aitrepreneur · YouTubeThe BEST local AI video generator is here! Minimax H3 tutorial
AI Search · YouTube
Frequently asked questions
It is a post-trained version of MiniMax H3, developed by fal on top of MiniMax's open-weight base model.
H3 Max ranks higher in image-to-video and follows prompts more closely; H3 offers 2K resolution and audio or video references.
From 71 to 451 credits per video, depending on duration, resolution. The exact cost is shown before you submit, and failed generations are refunded automatically.
480p and 768p; aspect ratios 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16. Clips run 5–15 seconds.
No. MiniMax H3 Max is developed by MiniMax and fal; Nano Banana is an independent service that runs it through a model provider.
Ready for MiniMax’s Best with H3 Max?
The top MiniMax model: frames, up to 9 references and native audio for polished clips.
- Clips up to 15s
- Native audio
- Failed outputs refunded