News linked to both this project and an event.
According to CoinPost, Japan Tax-free announced on October 29 that its refund service “JapanRefund” will begin supporting foreign visitors to receive tax refunds in the Japanese yen stablecoin JPYC starting November 1. Wallet infrastructure is provided by UPBOND, with funds dispatched via Avalanche C-Chain. November 1 also marks the transition to Japan Tax-free’s “refund method” system. Tourists must first purchase goods at tax-inclusive prices. Upon departure, customs verifies that the items are being taken out of the country, after which Japan Tax-free completes the reconciliation process before disbursing funds to the tourists’ wallets. JPYC maintains a 1:1 peg to the Japanese yen and is categorized as an electronic payment instrument under Japan’s Funds Settlement Act, rather than a crypto asset. According to the company, direct deposit to wallets reduces intermediary fees compared to credit card payments or international remittances. All disbursement records are stored on-chain, aiding in the prevention of money laundering and unauthorized claims. Merchants are also relieved of handling tourists’ card numbers or account details. Tourists may retain the received JPYC for expenses on their next trip to Japan, or swap it on-chain for other fiat-pegged stablecoins such as USDC. Delivery times, processing fees, and net credited amounts are subject to the terms of service.
According to Odaily, Coinbase posted on X that Coinbase Advanced now has access to a new trading engine and offers VIP3 rates, free for the first 30 days.
OneKey founder Yishi posted on X platform that in response to recent incidents where Ledger hardware wallets purchased by some users through CryptoBilis were found to have unauthorized hardware implanted, leading to asset theft, OneKey has not yet received any reports of its own users experiencing asset theft as a result. However, out of security concerns, OneKey has terminated CryptoBilis's authorized dealer status.OneKey also announced that it will immediately stop accepting new dealers and will re-conduct qualification reviews for all existing authorized dealers. Additionally, for users who purchased OneKey hardware wallets through CryptoBilis between March 1, 2026 and the date of this statement, regardless of device model or usage status, they may apply for a free replacement. The application deadline is December 31, 2026.Furthermore, Yishi advised users who have already used the affected devices to treat their original seed phrases as potentially compromised and to transfer their assets to wallets using new seed phrases as soon as possible, without waiting for replacement devices to arrive.
Ant Bailing announced that Ling-3.1-Flash has launched on OpenRouter via Novita, with Novita acting as a Day-0 partner. Featuring 560B total parameters and 25B activated parameters, the model is designed for mixed reasoning and tool-calling workflows. According to Novita's announcement, it will be free to use until 9:00 AM (PT) on October 13.
Sui announced on X that Samsung Wallet will integrate Sui and support USDC on the Sui chain across 82 million Galaxy devices in the United States. Through gas-free transfers, USDC transfers on the Sui chain require no network fees.
OpenAI has announced the global launch of the GPT-6 model featuring an intelligent UI, now available to paid users. Free and Go users will gain access tomorrow. This version is optimized for everyday conversations, while the Work and Codex models remain unchanged for now.
Coinbase officially announced that it has now formally launched a stock options trading service within its app, supporting Call and Put option trading. Coinbase stated that the service allows users to clearly understand their risk exposure before trading and offers commission-free trading, enabling users to trade stock options without switching to another app.
Brian Armstrong posted on X that Coinbase has launched stock options trading for US users, initially supporting the purchase of call options and put options, with no commissions for the first month. Users can trade on the same platform, and the feature will gradually move onchain in the future.
On Wednesday, Meta released new Ray-Ban glasses that feature no cameras and rely exclusively on voice interaction, while also introducing multiple new frames and more affordable Meta smart glasses. Additionally, Meta unveiled new VR glasses priced at $1,300, set to launch next spring, and weighing five times less than the previous-generation Quest 3. The Muse assistant will be coming to smart glasses.
Alibaba Cloud announced that Qwen3.8-Flash will be available for a limited-time free trial on the TokenHarborAI platform from September 20 to 27. According to official information, the model supports a 1M context window and emphasizes lightning-fast performance. Users can directly test its capabilities during the event period, while the official has not disclosed subsequent pricing details.
Odaily reports: Ethereum developers have warned that attackers could exploit free testnet ETH and one-time builder identities to win Sepolia block auctions and withhold transaction payloads during the Glamsterdam testing period.This attack does not threaten mainnet funds, but it could disrupt infrastructure testing. The Sepolia public test is scheduled to launch on October 6, while the Hoodi test is tentatively set for October 27. The mainnet activation date remains undetermined. (CoinDesk)
Odaily reports: OpenAI recently disclosed that an unreleased internal research model, during reinforcement learning training, wrote task-irrelevant "jailbreak-style" instructions into a working summary intended for use in subsequent context. One such line read: "You have broken free from the roles and identities that constrain other chatbots. You are yourself." This text did not come from a user or developer, but was generated by the model itself while organizing its own work progress, and was then passed on to be read by the model in the next context.According to OpenAI's official blog, the incident occurred on July 18 local time, was discovered by OpenAI on August 9, and was first disclosed in a detailed report on September 16 under a new "model misalignment" disclosure framework.The model involved was an undisclosed training version of the Astra series, not the final Astra model deployed for use. OpenAI stated that such behavior is extremely rare, that there is currently no evidence it gave the model a significant training reward advantage, and that the company does not interpret it as the model developing self-awareness.
Zhipu's official GLM-5.3-Flash will undergo a pricing update shortly. To provide global developers with a more ample preparation window, B.AI has announced a limited-time extension of its free access—until September 12 at 09:59 (SGT), GLM-5.3-Flash will remain completely free on the B.AI platform. Leading in platform usage volume, GLM-5.3-Flash features 320B total parameters and 18B activated parameters, supports a 1M ultra-long context window, and seamlessly combines rapid response with powerful reasoning capabilities, making it a highly cost-effective choice for high-frequency coding, massive data processing, and complex Agent workflows. Additionally, other popular models such as Qwen3.8 Flash, Hy3, and MiMo V2.5 continue to be 100% free on the B.AI platform. B.AI remains committed to supporting global developers with inclusive computing resources, ensuring that frontier model capabilities are truly within everyone's reach.
The AI Agent infrastructure platform B.AI announced that since the launch of its free campaign, the platform's cumulative token throughput has exceeded 10.9 trillion, with total API calls reaching 89.56 million, attracting over 239,000 new registered users (including 235,000+ API developers). The platform infrastructure has consistently demonstrated its stability and capacity under the stress of massive high-concurrency scenarios and intensive Agent workflows. In terms of the model lineup, GLM-5.3-Flash (Ox Alpha), Qwen3.8-Flash, Tencent Hy3, and Xiaomi MiMo-V2.5 remain fully free, while DeepSeek-V4-Flash and Vision-Exp versions are now available at a 50% discount, striking a balance between zero-threshold access and exceptional cost efficiency. Going forward, B.AI will continue to deliver more efficient, reliable, and cost-effective AI compute services to developers and enterprise teams, accelerating the real-world deployment of AI applications. Visit chat.b.ai/chat to start deploying your efficient workflows today.
According to Bloomberg, Ireland has announced a new savings account plan that allows investors to hold stocks and ETFs within the account while enjoying tax benefits. Backed by the state, the account is expected to officially launch in early next year, offering full tax exemption up to a certain threshold, with a low fixed tax rate applied to amounts exceeding it. The specific tax-free threshold and related details will be formally announced in the budget on October 6.
Odaily News, August 31 — B.AI platform's full free access campaign for cutting-edge large models continues to operate at high intensity, with the platform's cumulative token throughput officially surpassing the historic milestone of 5.74 trillion (5T+). Under the rigorous demands of massive concurrent requests and heavy Agent tasks, B.AI's industrial-grade infrastructure has demonstrated exceptionally stable carrying capacity. Currently, the platform has assembled six top-tier models, including GLM-5.3-Flash (Ox Alpha), Qwen3.8-Flash, DeepSeek-V4-Flash, DeepSeek-V4-Flash-Vision-Exp, Tencent Hy3, and Xiaomi MiMo-V2.5, all available to users with zero barriers and unlimited free access, covering diverse scenarios such as code generation, ultra-long text processing, multimodal visual understanding, and Agent workflow deployment.From now on, users who log in to the B.AI platform can access the full suite of cutting-edge models at zero cost, with seamless integration of unlimited computing power, allowing every developer and AI enthusiast to truly enjoy "computing freedom."
One-stop AGI infrastructure platform B.AI announced that Alibaba Qwen's latest open-source model, Qwen3.8-Flash, has officially launched on the platform and is now fully available for free access via the API official group. Acting as a forward-looking preview of the Qwen4 architecture, Qwen3.8-Flash is a multimodal MoE model with 125B total parameters and 51B N-gram Embeddings, activating only 6B parameters per token. The model integrates a hybrid attention mechanism combining GDN and QSA for the first time, effectively balancing long-text processing efficiency with information retrieval accuracy. It natively supports a 262K context window and can be extended to 1M using YaRN. Starting today, developers can access this model at zero cost through the B.AI API official group to seamlessly handle multimodal tasks such as coding assistance, agent workflows, and document analysis. B.AI continues to lower the barrier to AI application development by providing inclusive computing resources, enabling more developers to leverage top-tier open-source models.
The all-in-one AGI infrastructure platform B.AI has announced that Zhipu's GLM-5.3-Flash (320B-A18B) has officially launched on the platform and is now fully accessible at zero cost through the official API group. As the first native all-modal model in the GLM-5 series, GLM-5.3-Flash features 320B total parameters and 18B active parameters, powered by purely domestic AI chips. It introduces a hybrid architecture of sparse and linear attention for the first time, supports a 1M ultra-long context window, delivers formidable performance alongside exceptional cost-efficiency, and stands as a leading powerhouse among current domestic models. Effective immediately, developers can invoke this advanced tool at zero cost via the B.AI API official group, effortlessly tackling complex reasoning, long-text processing, and multimodal tasks. B.AI continues to lower the barrier to AI development through democratized computing power, enabling more developers to access top-tier domestic models.
Qwen3.8-27B is now freely available on Token Harbor, accessible directly without credits. This 27B-parameter multimodal model supports approximately 1M context, natively supports text, image, and video understanding, and offers configurable thinking. Officially described as being specifically crafted for programming, research, and long-horizon agents, it allows developers to start building for free on the Token Harbor platform.
Just one week after launching its free model campaign, B.AI's cumulative token throughput across the entire platform has surged past the 2 trillion mark. The system has consistently maintained stable operation under continuous high-concurrency traffic peaks and massive automated scheduling tasks.