Genmo servers crash – Hoshio

On October 22, Genmo, an AI-based platform for video production, announced a new model called the Mochi-1. This advanced and open-source model is able to produce high-quality videos using only simple texts.

In addition to Mochi-1, the company also launched an online environment where users can use Mochi-1’s features for free. The weights and structure of the Mochi-1 model are also publicly available on HuggingFace. But unfortunately, this server crashed within a few hours after its launch with a high volume of requests and it became difficult for users to access it.

However, the company quickly drew attention after the release of the Mochi-1 and admitted to the disruption to its website. In fact, the company attributed this problem to increased traffic. “We’re experiencing a lot of load,” Genmo announced on Platform X. Some users are experiencing errors, please bear with us during this time to increase capacity.”

“We’re back to normal fast! We added a bunch of GPUs to the environment and now everything is back to normal. You can use it now. You all welcomed our servers so much that they eventually melted. Thank you to everyone who waited for our capacity to increase. Please create and share your best texts and productions!”

The company claims that the Mochi-1 performs very well in creating realistic movements and following text commands.
Mochi-1 is an advanced video generation model that is trained using a propagation model with 10 billion parameters and is capable of generating videos at 30 frames per second. Genmo has released this model under the Apache 2.0 license, which provides suitable conditions for its use and development.

This model is similar to other proprietary models such as Runway’s Gen3 Alpha, Luma AI’s Dream Machine, Minimax’s Kuaishou and Hailuo. The Genmo team stated: “Our goal at Genmo is to showcase the capabilities of AI. “Mochi-1 is the first step in building simulators that can imagine anything (possible or impossible).”

In Mochi-1’s research preview, this model creates new possibilities in various fields and helps to advance video production techniques and explore new methods in research and development.

In terms of product development, Mochi-1 contributes to innovative applications in the fields of entertainment, advertising and education. The model also allows artists and producers to showcase their visions and expand their creative capabilities with AI-generated videos.

In addition, in the field of robotics, Mochi-1 helps generate artificial data for training artificial intelligence models in autonomous vehicles, robotics, and virtual environments, and responds to the needs of various industries.

Genmo has partnered with the best platforms to ensure that Mochi-1 is easily accessible to developers. Also, Mochi-1 enables it to be seamlessly integrated into applications using APIs from partner platforms, providing an easy way to exploit its capabilities in projects.