MiniMax H3 now available on AI Gateway
MiniMax H3 brings short 2K video generation to Vercel AI Gateway, with text, keyframe, and multimodal reference inputs. Reference and keyframe modes cannot be combined.
MiniMax H3 is available as **minimax/minimax-h3** on AI Gateway. It returns **2K MP4 videos lasting 5–15 seconds** from text, a starting image, first and last frames, or image, video, and audio references.
Use first-frame input when preserving an image’s aspect ratio matters; use references for subject or style guidance. Choose the input mode before building the request because keyframe and reference workflows use different constraints.
MiniMax H3 is available as **minimax/minimax-h3** on AI Gateway. It returns **2K MP4 videos lasting 5–15 seconds** from text, a starting image, first and last frames, or image, video, and audio references. Use first-frame input when preserving an image’s aspect ratio matters; use references for subject or style guidance. Choose the input mode before building the request because keyframe and reference workflows use different constraints. **Reference and keyframe modes are mutually exclusive**. The material provides no generation time, price, quality evaluation, or consistency measurements across the supported aspect ratios.
This expands AI Gateway from model routing into short 2K video generation with several distinct conditioning workflows. The mutually exclusive keyframe and reference modes make input-mode selection an application-level branch, not merely a model-name swap. With no price, latency, quality, or consistency evidence, it establishes availability and interface constraints rather than a preferred video route.