AI models on Craze
Every video, image, voice, music and sound model Craze can direct, with what each one makes and what it costs.
Every video, image, voice, music and sound model Craze can direct, with what each one makes and what it costs.
Starts at$0.053 per second of video
MiniMax's speed-optimized H3 variant, generating 1080p video with sound in seconds. Trades some accuracy for speed, ideal for rapid iteration.
H3 Max generates quickly, which changes how fast you can iterate.
H3 Max supports 1080p.
H3 Max's speed comes with a lower price tag.
Like base H3, Max understands text, image, video, and audio inputs as a unified context, not isolated inputs.
Craze creatively jams, researches and plans with you.
Craze generates anything imaginable. Video, audio, music, motion graphics and more.
Craze watches your footage and builds a final cut. With a real editor if you need it.
Every model here runs on every Craze plan, including Free, and Craze can use a different one for each shot.
| MiniMax H3 MaxMiniMax | Seedance 2.5ByteDance Seed Team | Seedance 2.0ByteDance Seed Team | Seedance 2.0 FastByteDance Seed Team | |
|---|---|---|---|---|
| Makes | Text + Image + Video + Audio to video | Text + Image + Video + Audio to video | Text + Image + Video + Audio to video | Text + Image + Video + Audio to video |
| Quality | Up to 2K | Up to 1080p | Up to 4K | Up to 720p |
| Length | Up to 15s per clip | Up to 30s per clip | Up to 15s per clip | Up to 15s per clip |
| Sound | Generated with the picture | Generated with the picture | Generated with the picture | Generated with the picture |
| From | $0.053 per second of video | $0.114 per second of video | $0.075 per second of video | $0.06 per second of video |
| Task | Inputs | Resolution | Length | Aspect ratios | Sound |
|---|---|---|---|---|---|
| Text to video | Text prompt | 480p, 768p, 1080p | 5 to 15 s | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | Always generated |
| Image to video | First frame, Last frame | 480p, 768p, 1080p | 5 to 15 s | Follows the input | Always generated |
| Reference to video | Reference, Video, Audio | 480p, 768p, 1080p | 5 to 15 s | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | Always generated |
| Scene insertion | Video, Reference | 480p, 768p | 5 to 13 s | Follows the input | Always generated |
| Video extension | Video | 480p, 768p, 1080p, 2k | 5 to 15 s | Follows the input | Always generated |
$0/ month
$19/ month
$59/ month
$199/ month
MiniMax H3 Max is a speed-optimized H3-family variant. When you need to iterate fast, generate multiple versions, or preview an idea before committing to a final render, H3 Max delivers in seconds. It trades some accuracy of base H3 for speed, but maintains multimodal input support and respectable resolution.
H3 Max is the speed option. When you need to iterate fast, generate multiple versions, or preview an idea before committing to a final render, H3 Max delivers quickly. That speed changes the entire creative workflow from waiting-and-watching to rapid exploration.
H3 Max generates quickly, which changes how fast you can iterate. Generate, watch, adjust, regenerate, repeat. Stay in the creative flow without waiting.
H3 Max supports 1080p. You can work at broadcast resolution while maintaining speed. This resolution is sufficient for most commercial and broadcast work, and the speed advantage makes iteration practical at scale.
H3 Max's speed comes with a lower price tag. That makes rapid iteration even more affordable. When speed is your priority, H3 Max delivers both on turnaround and on budget.
Like base H3, Max understands text, image, video, and audio inputs as a unified context, not isolated inputs. You can combine a character reference image, motion reference video, and audio mood track, and H3 Max synthesizes them together, just like base H3, but faster.
H3 Max handles recognizable public figures' likenesses. The combination of speed and face-handling capability makes it practical for celebrity and influencer work on tight timelines. The tradeoff is slightly more facial morphing than base H3, but the speed advantage often outweighs the accuracy cost.
When you show a client a first draft, get notes, and need to regenerate with refinements in minutes, H3 Max's speed is a competitive advantage. The cost savings also mean you can afford to generate multiple directions and let the client pick.
Fast iteration means quick feedback loops.
A good prompt is still important. "A close-up of an actor's face, confused expression, blue backlight" is better than "make it dramatic." Specific direction yields better results even with fast generation.
You pay per second. The rate scales with resolution: 480p is the cheapest tier, 768p is the middle tier, and 1080p is the highest cost. $0.053 per second of video Sound is always included. Use the calculator on this page to estimate a specific clip. The lower per-second cost compared to base H3 makes batch generation affordable.
When you need 10 different versions of the same shot to see which framing works, H3 Max's speed makes it practical. Generate dozens of variations in the time base H3 takes for a single render.
Generate a spot, show the client, get notes, regenerate with refinements. H3 Max's speed turns days into hours. The cost savings also mean you can afford multiple directions to show the client.
Generate multiple variations, multiple lengths, multiple color treatments, all in the time base H3 takes for a single render. Test different prompts, different references, and different settings without waiting.
Use H3 Max to generate placeholder footage while you're building the edit structure. Swap in higher-quality renders later if needed. The speed makes it practical to generate proxy footage for editorial before committing to final renders.
For TikTok and Instagram Reels, where speed of iteration matters more than perfect facial morphing, H3 Max's cost and speed work well. The turnaround supports live content strategy adjustments.
Base H3 is more accurate, especially with facial features. H3 Max is faster and cheaper, and supports 1080p. Choose H3 if accuracy is critical with recognizable faces. Choose H3 Max when iterating fast.
H3 Max Turbo is even faster and cheaper, but doesn't support video and audio inputs, only text and image. Choose H3 Max Turbo for text-to-video and image-to-video work where speed and cost matter most. Choose H3 Max when you need video reference motion or audio reference mood inputs alongside text and images.
Describe what you want the way you would brief an editor, and Craze plans the shots and writes the MiniMax H3 Max prompts for you. Every Craze plan, including Free, runs MiniMax H3 Max and every other media model, and Craze can use a different one for each job. Each video lands on an editable timeline with your cuts, captions and music, ready to export as MP4. Generations come out of your plan's monthly credits, from $0.053 per second of video, with top-ups when you need more.
Rapid preview and iteration: When you need 10 different versions of the same shot to see which framing works, H3 Max's speed makes it practical. Client reviews and rapid feedback loops: Generate a spot, show the client, get notes, regenerate with refinements. Batch generation and variations: Generate multiple variations, multiple lengths, multiple color treatments, all in the time base H3 takes for a single render. Live-action reference for editing: Use H3 Max to generate placeholder footage while you're building the edit structure.
MiniMax H3 Max is a speed-optimized H3-family variant. It generates up to 15-second video clips at 480p, 768p, or 1080p with native stereo sound.
Extremely fast. A 10-second video can generate in under 10 seconds. That speed makes it ideal for rapid iteration and preview generation.
Text prompts, reference images, reference videos, and audio tracks. H3 Max understands them together in a single prompt.
H3 Max trades some accuracy for speed. It can morph characters more than base H3, and observed detail is slightly lower, but it generates much faster and costs less.
Yes. It always generates stereo sound. You can silence it in the edit if you prefer to add your own audio.
$0.053 per second of video. The rate scales with resolution. Check the calculator on this page for an exact price.
every Craze plan, including Free. Every plan includes generation credits.
Yes. MiniMax H3 Max supports recognizable public figures' likenesses, though it may have slightly more morphing than base H3.
Yes. H3 Max accepts text, images, video, and audio inputs together in a single prompt.
ByteDance Seed Team
From $0.114 per second of video
ByteDance Seed Team
From $0.075 per second of video
ByteDance Seed Team
From $0.06 per second of video
