MiniMax tools

Create a Hailuo 3 (H3 / H3-Max) video task from a multimodal content array: one text item is required, with up to two optional first/last-frame images. The v2 API serves only the H3 family; Hailuo 2.x models use /v1/video_generation.

Create a Hailuo video generation task from a text prompt.

Animate a first-frame image into a Hailuo video, optionally guided by a prompt.

Hailuo 2.3 is a video generation model from MiniMax. It accepts text prompts and reference images as input and generates video output, supporting both text-to-video and image-to-video workflows.

Generate video with MiniMax-H3-Fast from a text prompt (text-to-video), from a starting and/or ending frame image (image-to-video), or guided by reference images, video and audio (reference-to-video), at 480P only, 5 to 15 seconds. H3-Fast is a low-cost tier that keeps multimodal reference support, but it is slower than H3-Max and H3-Max-Turbo and its clarity and detail vary between runs. Suited as a niche option to test against a specific requirement rather than a default; for speed use MiniMax-H3-Max-Turbo, for quality use MiniMax-H3.

Generate video with MiniMax-H3-Max-Turbo from a text prompt (text-to-video) or a starting and/or ending frame image (image-to-video). Outputs 5 to 15 seconds at 480P or 768P; reference images, video and audio are not supported. Turbo is the fastest tier in the H3 family and priced at half of H3-Max, so it is the right choice for drafts, batch exploration and rapid iteration. Text, faces and transitions are more prone to artifacts than on H3-Max, especially at 480P, so prefer 768P. Use MiniMax-H3-Max or MiniMax-H3 for final output.

Generate video with MiniMax-H3-Max from a text prompt (text-to-video), from a starting and/or ending frame image (image-to-video), or guided by reference images, video and audio (reference-to-video). Outputs 5 to 15 seconds at 480P or 768P; 2K is not available on this tier. H3-Max is the high-speed member of the H3 family: a 5-second clip renders in seconds, with strong overall quality at a lower per-second price than Standard H3, but weaker complex editing, reference fidelity and aesthetics. Suited for high-frequency creation where turnaround matters while still needing good image quality; use MiniMax-H3 for 2K or the best reference control.

Generate high-definition speech from up to 10,000 characters of text.

Generate fast, cost-efficient speech from up to 10,000 characters of text.

Generate music — a complete song from a text prompt and lyrics — with MiniMax music-3.0 (the current recommended model) or music-2.6. Provide lyrics for a vocal track, or request an instrumental; lyrics can also be written automatically from the prompt. Each run produces one generated song.