# Magnific will showcase MiniMax H3 at a San Francisco creator event > Source: > Published: 2026-08-13 01:15:08+00:00 [Magnific](https://www.magnific.com/company/about-us?ref=runtimewire), the creative AI platform founded by Javier "Javi" Lopez and Emilio Nicolas, will present MiniMax H3 at its San Francisco office on Aug. 13, 2026. The [event listing](https://luma.com/magnific-24eu?ref=runtimewire) describes a live demonstration covering cinematic generation, native audio, dialogue and editing. It names Magnific, MiniMax, Enrique Lopez, Victor Su-Ortiz and Leanna Ren as hosts. [MiniMax on X](https://x.com/MiniMax_AI/status/2087705790274752532?ref=runtimewire) The listing says Magnific has more than 110 million users and 1 million subscribers, giving the presentation a large potential audience of creative users. Magnific's [company page](https://www.magnific.com/company/about-us?ref=runtimewire) separately claims more than 100 million monthly visits. Neither source says H3 is available to those users as a product feature. ### Magnific is presenting H3 through a creator workflow The event, titled "Magnific Spotlight: MiniMax H3," is scheduled for 5:30 p.m. to 8:30 p.m. PDT. Magnific describes it as the first installment in a series intended to bring newly released AI models to creative users. Its agenda includes a user-project showcase, a discussion between MiniMax and Magnific, and a live workflow demonstration. That format puts H3 in front of people using AI for production rather than presenting it solely through model documentation or code. Magnific, which Freepik acquired in 2024, offers image, video, audio and 3D tools in one product, according to its [company page](https://www.magnific.com/company/about-us?ref=runtimewire). It also [markets access to models from multiple providers](https://www.magnific.com/ai?ref=runtimewire), making the platform a possible creator-facing distribution channel for H3 if the model becomes an available product feature. The [event listing](https://luma.com/magnific-24eu?ref=runtimewire) does not specify eligible subscription plans, usage prices, access restrictions, a rollout schedule or H3 adoption figures. It also does not explain whether the demonstration uses Magnific infrastructure, MiniMax's hosted services or a combination of the two. The available material establishes a scheduled presentation rather than general availability across Magnific's product. Yan Junjie founded MiniMax after spending more than six years at SenseTime, where he became a vice president and vice-head of its research institute. A mathematician by training, he earned his doctorate in artificial intelligence from the Institute of Automation at the Chinese Academy of Sciences and later conducted postdoctoral research at Tsinghua University. He is MiniMax's founder, chairman, CEO and CTO, according to the company's [management page](https://ir.minimax.io/corporate-information/management?ref=runtimewire). ### Magnific's H3 demo leaves infrastructure questions open MiniMax's [launch announcement](https://www.minimax.io/blog/minimax-h3?ref=runtimewire) says H3 accepts text, images, video and audio and can generate clips with native stereo audio at up to 2K resolution and 15 seconds. Its [model documentation](https://platform.minimax.io/docs/guides/models-intro?ref=runtimewire) lists clips from four to 15 seconds at 24 frames per second, with 768p and 2K output paths. MiniMax has published the [H3 repository and open-weight H3-Base checkpoints](https://github.com/MiniMax-AI/MiniMax-H3?ref=runtimewire), including separate variants for first-and-last-frame generation and multimodal reference generation. The repository says those checkpoints can produce 768p video and audio locally. It also says H3 jointly predicts video and audio representations before separate decoders produce the final video and 32 kHz stereo sound. The repository distinguishes those local checkpoints from two hosted parts of the system. H3-Context-IR processes complex multimodal instructions through MiniMax's API, while H3-Regenerate-2K uses a 768p result and the original context to regenerate video at 2K. Neither component is included in the open repository, according to the [H3 project page](https://github.com/MiniMax-AI/MiniMax-H3?ref=runtimewire). Magnific's event materials do not identify which of those components underpin its live workflow. That leaves open whether the demonstration runs H3-Base locally, calls MiniMax's hosted context and 2K services, or combines local and API processing. The answer would determine how much of a future Magnific offering could run on its own infrastructure. MiniMax says it was founded in 2022 and provides models and products spanning language, video, audio and music. Its [company profile](https://ir.minimax.io/corporate-information/company-profile?ref=runtimewire) reports more than 212 million individual users across over 200 countries and regions, plus more than 100,000 enterprises and developers across more than 100 countries. The Aug. 13 presentation should show how Magnific places H3's generation, sound, dialogue and editing controls inside a workflow aimed at creative users. Product availability, pricing, plan access and the workflow's dependence on MiniMax's hosted services remain unspecified in the event materials.