Qubic’s science team made 155,192 evolved AI brains public. They were built for the ARC-AGI-3 test. They run on regular CPUs with no GPUs and no LLMs.
Key points
Released: 155,192 evolved AI brains
Built for: ARC-AGI-3 test
Hardware: run on regular CPUs, no GPUs, no LLMs
Sources
GEODNET network enables cm-level RTK for parking spaces
GEODNET's network of 22K+ stations across 176 countries provides RTK accuracy of a few centimeters, enough to identify a specific parking space. A parking space is about 2.5 meters (8 to 9 feet) wide, and regular GPS can be off by several meters.
Key points
Parking space width: about 2.5 meters (8 to 9 feet)
RTK accuracy: a few centimeters, identifies space, row, and side of line
Sources
Firelight first layer of protection on LidoFinance EarnETH vaults 5.7% $12.1M
Firelight is now the first layer of protection on LidoFinance's EarnETH and EarnUSD vaults. For the first 30-day period coverage is capped at 5.7% of EarnETH’s total value locked, about $12.1 million.
Key points
Firelight provides the first layer of protection on LidoFinance's EarnETH and EarnUSD vaults.
Coverage is capped at 5.7% of EarnETH’s total value locked, roughly $12.1 million for the first 30 days.
Sources
LidoFinance EarnETH Firelight coverage 5.7% (~$12.1M)
LidoFinance adds Firelight as the first layer of protection on its EarnETH and EarnUSD vaults, with coverage capped at 5.7% of EarnETH’s total value locked (~$12.1M) for the first 30 days. Initial 30-day limits are about $12.1M for EarnETH and $4M for EarnUSD, and payouts depend on available collateral and can be zero.
Key points
Coverage capped at 5.7% of EarnETH’s total value locked, approximately $12.1M.
EarnUSD vault initial 30-day limit is about $4M; payouts depend on collateral and may be zero.
Sources
MAG 7 ARE DOWN $800 BILLION so far today
MAG 7 ARE DOWN $800 BILLION so far today, with Nvidia down $320B and Tesla down $144B. The post lists losses for Microsoft, Alphabet, Amazon, Meta, and Apple as well.
Key points
Nvidia loss: -$320B
Tesla loss: -$144B
Sources
PAGI Network launches new coordination, verification, reward system
PAGI Network is introduced as a new way to coordinate, verify, and reward the work between raw data and model training for Physical AI. It addresses the data problem in Physical AI by offering a structured network.
Key points
Problem: Physical AI has a data problem.
Solution: PAGI Network coordinates, verifies, and rewards work between raw data and model training.
Sources
Flare Data Connector enables FXRP minting and staking
Flare Data Connector verifies XRP transactions, enabling FAssets to mint FXRP on Flare Networks. FXRP can be staked in Firelight to provide capital backing DeFi cover, protecting LidoFinance Earn users on Ethereum.
Key points
Data Connector verifies XRP transactions on the XRPL.
FXRP can be staked in Firelight to provide capital backing DeFi cover for LidoFinance Earn users.
Sources
Kōhai Free launch – first 1,000 users free
Kōhai is live and offers free access to its AI team for the first 1,000 users. No credit card is required and the free tier includes monthly credit refreshes. Paid plans start at $20 / month for additional credits.
Key points
Free tier limited to the first 1,000 users, no card needed
Paid plans start at $20 per month with more credits
Sources
- Original post
- Kōhai Free: limited to the first 1,000 users - Linked page