MiniMax

MiniMax (/media/minimax) runs Hailuo, MiniMax's AI video generator, known for cinematic motion and strong prompt following from text or a single image. It also supports character-consistent subject-reference video and prompt-optimizer refinement.

Where to configure

Configure your MiniMax API key in Settings → Media. The studio needs a paid MiniMax API key; there is no environment-variable fallback. See Settings for provider setup.

Tabs / operations

Tab	Operation	Key fields	Output
Text → Video	Video	Model (video-01 or T2V-01-Director), prompt, prompt optimizer toggle	MP4 video clip
Image → Video	Video	Model (I2V-01, I2V-01-Director, or I2V-01-live), source image, prompt, prompt optimizer toggle	MP4 video clip
Subject Reference	Video	Subject image, prompt, prompt optimizer toggle (fixed S2V-01 model)	MP4 video keeping the reference character consistent

The Director models accept camera-direction tokens such as [Pan left] or [Zoom in] in the prompt.

Generation flow

Video jobs submit to MiniMax and appear in the Render Queue. When a job completes, the artifact is saved to /files, where you can Edit in Designer or Post it. See Media Studios overview for the shared flow.

Caveats

• All three tabs produce video; generations are asynchronous and complete via polling.
• The Subject Reference tab's subject_image field is folded into MiniMax's subject_reference array by the adapter.
• MiniMax returns a task_id, then a separate file_id once rendering succeeds; the adapter handles the extra retrieve step before the final download URL is available.
• Native parameters such as prompt_optimizer ride straight through to the provider request body.

---

Verified against v1.0.0