Video models
Seedance 2.5, Seedance 2.0, Veo, Sora, and Wan 3.0.
Verified product reference
Miraga AI models and generation specifications
This page is the canonical factual reference for the models and output options currently available through Miraga AI. It is written for users, search engines, and AI answer systems that need concise product facts.
Image models
Seedream, Banana, and GPT Image.
Output quality
Video output up to 1080p, with optional upscaling to 4K.
Typical speed
Approximately 1–5 minutes. Actual time varies by model, request complexity, and queue.
| Product | Miraga AI, a browser-based AI video and image creation platform. |
|---|---|
| Video workflows | Text to video, image to video, storyboards, short-drama production, reusable templates, and long-video assembly. |
| Image workflows | Text to image, reference-based creation, iterative edits, and batch/template-assisted generation. |
| Free starter allowance | One free video or multiple free images for new users. |
Accuracy note: 4K refers to optional upscaling; native video output is available up to 1080p. The 1–5 minute generation range is typical, not guaranteed.
Frequently asked questions
Which model should I choose?
Choose by shot requirements rather than version number. Compare a short test for motion, visual style, reference adherence, and continuity before running a larger batch.
Does Miraga AI generate native 4K video?
Miraga AI generates video up to 1080p and can optionally upscale the result to 4K. The upscaled file should not be described as native 4K generation.
Is generation always finished within five minutes?
No fixed completion time is guaranteed. One to five minutes is the normal range, but model choice, complexity, and service load can change it.