Hailuo 3 API

What is Hailuo 3 API

Hailuo 03 (MiniMax H3) API is a service that generates native 2K video clips with synchronized stereo audio from prompts and reference media.

How to use Hailuo 3 API

To use the Hailuo 03 API, send a POST request to /v2/videos/generate with your API key. The request should include a prompt, the model ID hailuo-03, and optional parameters within hailuoParams such as resolution, image, end_image, base_video, reference_images, reference_videos, and reference_audios. The API returns a jobId that can be polled for status, or a webhook_url can be provided to receive notifications when the video is ready.

Features of Hailuo 3 API

• Native 2K Video with Stereo Audio: Generates video clips at 768p or native 2K resolution with synchronized stereo audio in a single pass.
• Omni Reference Input: Accepts up to nine images, three video clips, and three audio samples as references to maintain consistency in subject, setting, motion, and voice.
• First and Last Frame Control: Allows users to specify the opening and closing frames for motion generation or to animate a still image.
• Flexible Durations: Supports clip lengths from 4 to 15 seconds, billed per second.
• Per-Second Pricing: Billing is based on the duration of the generated clip, with additional charges for reference videos and extra reference images.
• Unified API: Accessible through the same endpoint and API key as other Apiframe models.
• Instruction-Based Edits: Enables modifications like changing backgrounds or wardrobes through descriptive prompts.

Use Cases of Hailuo 3 API

• High-Detail Hero Footage: Creating clips with fine textures suitable for large screens or crops.
• Sound-Designed Clips: Generating clips with spatially placed ambience and effects.
• Reference-Driven Consistency: Maintaining character or visual identity across multiple clips using reference inputs.
• Instruction-Based Editing: Modifying existing shots based on textual descriptions.

Pricing

The pricing for Hailuo 03 is 14–22 credits per second of output, approximately $0.14–$0.22 per second. A 9-second clip would cost between 126–198 credits (≈$1.26–$1.98). Reference videos are billed at the same per-second rate, and reference images beyond the first five incur a small flat charge each. Billing is per second of output, and failed jobs are refunded. Volume discounts are available.

FAQ

• What is Hailuo 03? Hailuo 03 is the model ID for MiniMax H3, a multimodal video model that generates 768p or native 2K clips with synchronized audio from text, frames, or reference media.
• What inputs does it accept? It accepts a text prompt alone, a first and optional last frame, or a set of reference images, videos, and audio clips. Frame and reference inputs cannot be combined in the same request.
• Does it generate audio? Yes, audio is produced in the same pass as the video and is muxed into the MP4 file, including dialogue, sound effects, and ambience. An audio reference can be supplied to guide the voice.
• What resolutions and durations are supported? Supported resolutions are 768p or native 2K. Durations can be any whole second from 4 to 15 seconds. The default is 2K. A 768p clip can be used as base_video to regenerate it at 2K.
• How is it billed? Billing is per second of output. Reference videos are billed at the same per-second rate, and reference images beyond the first five have an additional flat charge. Only successful generations are billed.
• Is the model open weight? MiniMax has committed to releasing H3 weights, but they are not yet published. The API functionality remains the same regardless of weight availability.