Grass

Description

The value of the network shows up in how it’s being used. Here’s how customers are already working with Grass: The customers below remain unnamed, since their work with Grass gives them a unique edge in AI where every advantage matters. • Customer A needed an internet-scale corpus of video/audio data. Grass delivered 100M+ hours via automated pipelines, enabling training for some of the most powerful models in the market. • Customer B required text-to-speech data where the same speakers switch languages, a current gap in the industry. Grass built pipelines to detect multilingual audio tailored to their needs, delivering curated datasets that power dubbing and translation. • Customer C required curated video data to fine-tune smaller models. Grass provided access to billions of indexed videos with metadata filtering, allowing the client to fine-tune smaller models that achieve large-model performance for specific use cases at a fraction of the cost. 2025 has been about pushing the limits of what the network can accomplish. None of it would exist without the users of Grass, and we look forward to continuing to build a more open internet together.