DeepSeek Breaks the "Impossible Trinity" of Large Models: Stronger, Faster, and Cheaper

Stablecoin payments startup Raycash, spun off from privacy protocol Zama, has launched on the Ethereum mainnet and is currently in private beta. Raycash states that over 75,000 people joined its waitlist ahead of the launch; access will be rolled out in batches during private testing, with full public access to follow. Built on the Zama protocol, Raycash uses fully homomorphic encryption (FHE) to encrypt on-chain balances and transaction amounts. Users can wrap stablecoins 1:1 into tokens compliant with the ERC-7984 confidential token standard for payments and receipts, and earn yields via Morpho, with transaction amounts and balances kept private. The underlying stablecoins can be retrieved via permissionless unwrapping. The app initially supports USDC, allowing deposits via bank transfers, debit cards or stablecoins, and withdrawals to bank accounts. Ethereum gas fees are handled by the platform in the background, so users do not need to hold ETH. Raycash adds that its wallet is non-custodial; confidential tokens have no built-in freeze or seizure functionality at the token level, and users can grant view permissions to specific parties like accountants or auditors. Additionally, a Raycard for privacy-focused spending is in development.

Deutsche Bank and Ant Digital Technologies jointly published the whitepaper *Agentic Payments in Wholesale Finance* on November 9, exploring the application potential of AI Agents in areas including institutional fund management, cross-border payments, and trade finance, along with corresponding governance and risk control frameworks. The whitepaper points out that AI Agents could autonomously predict liquidity gaps, evaluate fund allocation and foreign exchange transaction plans, coordinate cross-border payment routes, process trade finance documents, identify data discrepancies, and generate payment instructions. Such applications are expected to enhance the automation level and operational efficiency of financial institutions’ complex businesses. For risk control, the whitepaper advocates the "Bounded Agentic Execution" model instead of granting AI unlimited autonomy, and proposes a reference architecture consisting of three layers: intent and orchestration, control and authorization, and settlement execution. Core safeguards include clear and revocable permission boundaries, verifiable agent identities, and the Know Your Agent (KYA) verification mechanism. It also lists Ant Digital Technologies’ self-developed agent platform Anvita Flow as a practical application case.

Dongcha Beating AI News reports that ByteDance has released its next-generation image model Seedream 5.5, which supports up to 100 rounds of continuous edits on a single image and can directly generate transparent background assets. ByteDance showcased a demo of 100 rounds of continuous editing: centered on the same character, each round builds on the previous image to alter clothing, glasses, rings, and desktop items. After repeated edits, the character retains its original features, and the model can accurately adjust small details like rings and frames. The company states the new version keeps unmodified areas pixel-consistent, avoiding past issues of increasing distortion and color bias with edits. Another new feature is native RGBA transparent asset generation: for example, when generating a snowy owl with spread wings, the result comes with a transparent background, even handling tiny gaps between feathers. These assets—including stickers, text graphics, and product materials—can be directly added to posters or design drafts, eliminating manual matting steps. Generation quality has also improved: the new version reduces issues like overly smooth skin, greasy visuals, and local overexposure in portraits, and can generate more complex Chinese text and dense small characters. According to ByteDance’s internal tests, the severe text corruption rate dropped to 0.4%, and the minor text corruption rate is 1.7%. The usability of results for film and office scenarios is over twice that of the previous Seedream 5.0 Pro. Seedream 5.5 is now rolling out on Jidream, Doubao, Doubao Work, and Volcano Engine, with a Pro version set to launch later.

Blockchain security firm Slow Mist issued a statement noting that the Ankr cross-chain bridge attack resulted in approximately $304,500 in losses. The attacker leveraged a single legitimate Optimism transaction to submit five forged cross-chain receipts, modifying the recipient, token, and withdrawal amount. All five receipts carried valid signatures from the cross-chain bridge’s trusted consensus signers; the withdrawal function released assets without independently verifying the actual event on the source chain. The source of the unauthorized signatures remains under investigation, and it has not been confirmed that the signers were compromised.

According to market data from BIT Stock Platform, Nvidia (NVDA.O) fell more than 4% in pre-market trading on US stocks.

Japan’s digital asset issuance and management platform Progmat has signed a memorandum of understanding (MOU) with U.S.-based Digital Asset Holdings, planning to migrate the trading and settlement of real-world assets (RWA) including Japanese government bonds (JGBs) onto the blockchain via the institutional-grade Canton Network. The two parties’ first phase will focus on integrating Progmat’s platform with the Canton Network, developing tokenized JGB products (tJGB) that retain the existing legal attributes of Japanese government bonds, and linking with existing securities transfer ledger systems. While upholding the legal validity of JGB right transfers and settlement finality, they will explore enabling 24/7 on-chain settlement. The Digital Asset Creation Consortium (DCC) led by Progmat has established a dedicated working group, and plans to launch the second phase of live testing in November. This phase will test the issuance and management of test assets including tJGB, tokenized deposits, and stablecoins on the Canton Network, and gradually connect domestic and international repo trading and collateral platforms. Earlier, MUFG and its three subsidiary financial institutions also announced their participation in the cooperation, to verify the full on-chain process of delivery versus payment (DvP) for government bonds and digital assets, as well as repo transactions.