@ImpactfulX@grass@grassfdn There’s no leaked information. Leaked info is misinformation. Please only refer to Grass official and not random pages posting fake info
@dutadrops Nothing about this post is true so I don’t know why you’re posting it. The tiers in your dashboard have nothing to do with the airdrop. Those tiers have been there since stage 1 and just give bonus points. Stop intentionally misinforming people lol
Join us on Discord for the first FliteGrid AMA!
CEO @gjordan will be answering your questions next Thursday, June 4th at 1:00pm EST.
Any questions you have can be posted in the # ama-questions channel, just let us know if you need an invite 👇
@Rahulweb3gyan@grass Is not about how much time you spend. In fact it doesn’t require you to spend any time to run the app since it just runs in the background while you use your device like normal. It’s all about the points you have
we're now working with 20 of the most exciting organizations in the AI industry, across frontier labs, public corps, startups, research institutions, and nonprofits.
@grass unlocks the public web - if you need billions of rows of training data, reach out to us at https://t.co/mTkvuQ5XtU
2025 has been about shipping. Here’s where Grass stands today:
Network
• 8.3M users have run an active node in the last 30 days.
• 1M+ concurrent user connections.
• Currently retrieving 3PB per day, driven by customer demand.
Infrastructure
• Storage system now running at 10% of hyperscaler cost with 2x the throughput capacity.
• Tape-based archival solution, built for 30-year durability, is 50-60% cheaper than hyperscalers’ cold storage tiers. Compute stack operating at up to 95% lower cost than cloud providers. These solutions provide full retrieval control, avoid vendor lock-in and also eliminate egress fees.
• Grasshopper hardware launching soon, designed to offer consistent uptime and throughput, strengthening reliability across the network.
Products
• Dataset catalog has grown to 4B+ multimodal files, 300M+ hours of audio, and trillions of tokens of text, with custom enrichments available across all datasets.
• Pre-configured industry datasets (finance, retail, staffing, AI) are being listed in data marketplaces for direct access.
• ClipTagger live on @inference_net and HuggingFace, outperforming larger models like Claude 4 and GPT-4.1 on standard annotation metrics like ROUGE and BLEU.
• Video search beta enables transcript, semantic, and keyframe-level discovery across billions of videos, currently in testing with enterprise and research partners.
• High-volume APIs (including SERP and web video retrieval) in development.
With these structures in place and continuously growing, Grass delivers the retrieval and processing of web-scale data with an efficiency, cost-effectiveness and speed that surpasses what centralized providers can deliver.
despite it being the best keyframe annotation model in the world at its price point; for me, the coolest part is that petabytes of data were collected via a network of millions of real people's devices, used to train a model, and then deployed on a distributed network of GPUs
@GrindLocal63327 Yes and that will be announced when the time comes! Last epoch is not known and is dependent on progress and development of these things!