Hailuo 2.3 AI Video Generator
Create 768p videos with Hailuo 2.3 from text or one start image. Best for physics-aware motion, action shots, lighting shifts, and 6/10s drafts.
Examples
Related Models
Model Guide
Hailuo 2.3
MiniMax Hailuo 2.3 is a video model for physics-aware 768p short clips. Official documentation positions it around stronger physical motion, 2.5x efficiency, and an 85% complex-instruction response rate; in Epochal, the public workflows now support prompt-led text-to-video and first-frame image-to-video with 6 or 10 second duration presets.
Start
Choose prompt-led or first-frame video
Write executable motion
For text-to-video, describe the subject, action, force, speed change, camera path, and visible feedback such as splashes, dust, cloth, or reflections.
Anchor image-to-video with one clean frame
For image-to-video, upload one clear start image with the subject and composition intact, then use the prompt only to guide the movement that follows.
Use duration as a decision gate
Start with 6 seconds when testing direction. Move to 10 seconds only after the core action and camera behavior are close.
Decision
How parameters change the result
mode
Impact: Standard is the default and exposes 6/10 second duration presets. Pro uses the dedicated Pro endpoint at a fixed per-run cost and does not expose duration.
Recommendation: Start with Standard for duration-controlled drafts. Switch to Pro when you want the Pro endpoint and accept fixed per-run pricing without duration control.
image
Impact: Image-to-video requires one start image. The image controls subject, composition, and the first-frame visual style.
Recommendation: Use a clear JPEG, PNG, BMP, or WEBP. Avoid cropped subjects and use a frame that already represents the shot you want to animate.
duration
Impact: Both workflows support 6 or 10 seconds. Shorter clips validate composition and motion stability; longer clips give the action more room to develop.
Recommendation: Use 6 seconds for the first pass. Switch to 10 seconds after the subject, motion, and camera direction are close.
prompt
Impact: Prompt clarity matters for action chains, camera movement, subject relationships, and physical feedback. In image-to-video, the prompt can be optional but still helps direct motion.
Recommendation: Avoid style-only prompts. Write what happens, how the camera moves, and which visual details must remain stable.
Cost
Cost and runtime expectations
| Workflow | Duration | Estimated credits |
|---|---|---|
| Standard text to video | 6s | 84 |
| Standard text to video | 10s | 168 |
| Standard image to video | 6s | 84 |
| Pro text/image to video | no duration option | 147 |
Method
How to judge the first result
Start with the right input: Use text-to-video when the idea is still prompt-led. Use image-to-video when you already have a character, product, scene, or key frame that should anchor the clip. Standard shows duration control. Pro follows the upstream Pro schema and charges per run without a duration option.
Then inspect continuity: After generation, check subject stability, physical feedback, action continuity, lighting changes, and the camera path before extending duration or rewriting the prompt.
- • Use precise verbs for the beginning, change, and end of the action.
- • For image-to-video, make sure the start image is clear and not heavily cropped.
- • When comparing 6s and 10s outputs, change only duration first.
FAQ
FAQ
- What workflows are public for Hailuo 2.3 on Epochal?
- The public page supports text-to-video and image-to-video. Image-to-video requires one start image; text-to-video requires a prompt.
- What resolution does Hailuo 2.3 generate?
- This integration follows the Hailuo 2.3 Standard API documentation, which describes 768p video output.
- How should I choose between text-to-video and image-to-video?
- Use text-to-video to explore a scene from words. Use image-to-video when a specific subject, style, product, character, or composition should be preserved from the first frame.
- How much does Hailuo 2.3 cost?
- The current estimate is 84 credits for 6s T2V, 168 credits for 10s T2V, 84 credits for 6s I2V, and 168 credits for 10s I2V.