AI Models We Use

Video generation

Every neon AI video on this site is generated with MiniMax Hailuo 3 — model family name H3 (vendor: MiniMax, model identifier minimax-h3). H3 produces 4–15 second clips at 720p or 768p with an accompanying stereo audio track; our default output is 8 seconds in 9:16, 16:9 or 1:1.

The model is accessed through our model provider, kie.ai (API identifiers minimax-h3/text-to-video, minimax-h3/image-to-video and minimax-h3/reference-to-video). No other video model is used for rendered clips.

Image generation

Reference stills and concept frames are created with OpenAI GPT-4o Image (model identifier 4o-image-api), also accessed through kie.ai.

What we do not do

We do not train or fine-tune any model on your prompts or uploads. Prompts and reference images are sent to the model provider only to render your video, and are not sold or shared for advertising.

Model changes

If the underlying model changes, this page is updated before the change reaches production. The model in use is also shown in the MODEL selector on the studio panel of the home page.

Contact

Questions about model disclosure: [email protected].