Justin Sun: AI agents that utilize cryptocurrency are expected to become mainstream, with banks gradually exiting the market.

According to Arkham’s monitoring, trader CryptoCurb bought ZCAT in 38 separate transactions in early September, investing a total of around $19,600. At that time, the token’s market cap stood at approximately $2.17 million. The trader’s position has since risen roughly 15 times in value. They have now cashed out $49,900, while still holding ZCAT worth $259,400, bringing their total holdings to about $309,300; this sale represents a quarter of the total amount they purchased.

Sui announced in a post that it has launched Circle CCTP V2, enabling native USDC transfers between Sui and 29 other supported blockchains. Minting rights are held by Circle, and the USDC supply on Sui is directly sourced from the issuer.

According to Bitget data, WTI crude oil futures surged 3.00% on the day, currently trading at $90.93 per barrel. On the news front, the Pentagon has ordered U.S. Central Command (CENTCOM) to complete preparations for resuming major military operations against Iran, but Trump has not made a final decision nor set a specific strike date. U.S. and Israeli officials said that if the operation is restarted, it could be scheduled before the U.S. midterm elections on November 3, and even earlier than Israel’s general election on October 27.

Liquid Capital founder Jack Yi posted that over the past month and a half, he has repeatedly tweeted six times that Bitcoin would pull back at the $86,000 level. Bitcoin has been consolidating sideways for the last half month, even seeing fake breakouts to lure bullish traders, but his team has stuck to the pullback call. Since accurately calling the end of Bitcoin’s rally in May, and repeatedly advising to buy the dip in July and August, their efforts have once again been validated. Investing and trading are highly falsifiable endeavors, but in the crypto space, if you lack the ability to operate projects and exchanges, only by continuously improving and honing your investment and trading skills can you have a chance at earning returns.

Lending protocol Loopscale has disclosed details of its merger with Solana ecosystem DEX Orca in official documentation. According to the docs: - Post-merger, Loopscale and Orca will continue to operate as products under Formation. - Loopscale’s point conversion will take place later this year, with specific details to be announced soon. - Points will continue to accrue until the redemption event starts. - Loopscale will not issue its own token; ORCA is the only token in the Formation ecosystem. BlockBeats previously reported that Solana ecosystem DEX Orca and lending protocol Loopscale announced their merger, with the new entity renamed Formation. Luke Truitt, co-founder of Loopscale, will serve as CEO, while Christopher Montagano of Orca will hold the positions of Chief Strategy and Legal Officer.

Beating AI Express (Dongcha) News: a16z-backed local AI assistant Underdog has launched Saluki 27B, a model built on Alibaba's Qwen3.8-27B. The model uses approximately 2-bit quantization, cutting its size from ~54GB (full-precision version) to 7.89GB. According to officials, it can run on laptops with 16GB of RAM and supports deployment via existing tools like llama.cpp. The model weights are open-sourced under the Apache 2.0 license. Saluki leverages quantization work from ISTA-DASLab (Austrian Institute of Science and Technology). The lab previously compressed Qwen3.8-27B to ~8.4GB; Underdog further reduced the model size while focusing on optimizing tool call performance. In 120 tool call tasks selected from BFCL, Saluki answered 88 correctly, the ISTA-quantized version got 76 right, and the 54GB original version scored 84. In 50 real GitHub issue-fixing tasks, Saluki resolved 30, while the original version fixed 33. Officials noted that Saluki retains an average of 96% of the original model's performance across 9 benchmarks, though performance varies significantly across different tasks. Its scores on math competitions and complex reasoning tasks only reach ~83% and 85% of the original, respectively.