Decart

Decart's official company timeline reports that the 2024 Oasis real-time AI-generated game launch reached one million users in 72 hours. This historical browser-game usage observation is distinct from the previously recorded Glint Android download threshold.

The Google Play listing for Decart's real-time generative entertainment app shows a 10K+ download threshold and an August 20, 2026 update, providing an observable consumer distribution indicator.

Decart has demonstrated and productized real-time generative world/video systems, while broad enterprise deployment remains relatively early.

Decart research reports Lucy 3 inference at 101 FPS on AWS Trainium and optimization across TPU and GPU hardware. These company-reported implementation measurements add a specific systems research result beyond the existing general model description.

Decart develops real-time generative world and video models with differentiated inference and model architecture work.

Decart develops real-time generative world and video models with differentiated inference and model architecture work.

Decart's official company timeline reports a latest US$300 million funding round involving Radical Ventures and more than US$450 million in total funding.

Backblaze's Decart case study describes Decart's team testing usage patterns and operating multi-cloud model-training infrastructure with petabyte-scale dataset transfers to multiple GPU clusters. It reports growth to 16PB in 90 days and quotes Decart's CEO on the operational storage reliability. This completed engineering/infrastructure execution is distinct from the previously recorded financing round.

Decart has built a specialized model research and systems organization capable of training and serving compute-intensive generative models.

Decart's realtime video-editing documentation demonstrates WebRTC stream integration, remote-stream callbacks, connection-change/error handlers and explicit disconnect cleanup. These specific lifecycle mechanisms support an analyst inference of portable realtime media and asynchronous SDK integration skills.

Generative modeling and inference systems skills transfer broadly, while world-model specialization narrows demonstrated portability.

Generative modeling and inference systems skills transfer broadly, while world-model specialization narrows demonstrated portability.