source

According to Rest of World, Alibaba launched the open-source model platform ModelScope in 2022, and OSChina launched MoArk in 2023, providing model hosting, testing, and customization services for Chinese developers behind firewalls. In 2023, Chinese regulators blocked Hugging Face without disclosing specific reasons; reports indicate that the government simultaneously allowed artificial intelligence laboratories to bypass the blockade to share Chinese models globally. ModelScope currently hosts over 170,000 models and stated in March that it has 170,000 models and 250 million users; MoArk hosts about 20,000 commonly used models, both still below Hugging Face, which hosts over 3 million open models.OSChina CEO Xu Yong stated that not everyone can always use a VPN, and China needs to establish an independent artificial intelligence ecosystem that serves Chinese users, claiming that China is developing an independent ecosystem faster than during the internet era. He also mentioned that MoArk has deployed engineering teams to ensure that models can run on different types of Chinese chips, making it their mission to ensure mainstream open models are compatible with Chinese chips. Both platforms offer some services for free, with additional computing power and premium features charged; ModelScope also hosts university hackathons and has opened co-working and event spaces for AI entrepreneurs in Hangzhou.Headquartered in New York, Hugging Face is a major platform for sharing open models and datasets. In September, Nvidia announced its acquisition of the company for $12.9 billion and noted in regulatory filings that regulators might prohibit sharing Chinese models on the site.

The artificial intelligence company DeepSeek has officially open-sourced infrastructure components for the Huawei Ascend computing platform, covering the TileLang high-level language compilation tool, computing libraries, and distributed communication libraries, corresponding to the previously open-sourced components for the NVIDIA platform. TileLang aims to provide a general-purpose, simpler programming language that can achieve the hardware performance limits, improving development efficiency and simplifying logic compared to CUDA, while its programming model can also leverage chip features.The TileLang route was first validated on the NVIDIA platform and has already supported the implementation of most operators in the training of the DeepSeek V4 series models. The open-sourced Ascend version encapsulates the underlying instructions of the Ascend C, providing a high-level programming approach without sacrificing hardware performance. Currently, every TileLang operator used in DeepSeek's training has a corresponding high-performance implementation on Ascend.The components open-sourced at the same time also include DeepGEMM, DeepEP, TileKernels, FlashMLA, and DeepSelect, which are used for general matrix operations, large-scale cross-device communication, conventional vector calculations and memory access, long-context sparse attention, and data filtering, respectively. DeepSeek claims that in multiple key tests, the related computing and communication performance has approached hardware limits; during the R&D process, the Huawei team provided support, and both parties collaborated to advance the 128-card supernode solution based on Ascend 950, with deep optimizations made for computing and communication.

NVIDIA (NVDA.O) launched an open intelligent agent security platform, ensuring the safety of intelligent agents throughout the entire process from testing to deployment.

According to a survey by the Financial Times, open-weight models are spreading from technology companies to traditional American enterprises. AlphaSense data shows that in August and September of this year, American companies mentioned open weights or open-source models approximately six times more than the same period last year during earnings calls and investor meetings. PNC Financial, logistics company CH Robinson, and Siemens have recently publicly discussed related applications.Tinder's annual AI spending has increased from about $1 million in January of this year to about $10 million in July, and it has begun to delegate some routine requests to open-weight models. AT&T stated that currently about 40% of AI tasks have been assigned to open models, with plans to increase this to about 70%. Its internal AI system processes approximately 45 billion tokens daily, using lower-cost models for simple tasks while continuing to use cutting-edge models like OpenAI and Anthropic for complex tasks.

Ethereum co-founder Vitalik Buterin announced today that the original novel "Snowmoon" has been completed, consisting of 32 chapters.The novel unfolds in a speculative science fiction setting, depicting a world that heavily relies on cryptography and decentralized governance. The story focuses on the ideological conflicts between societies such as Veridia and the Arctic Empire, involving elements like zero-knowledge proofs, second voting, privacy technologies, and content-oriented tax mechanisms, integrating themes of mechanism design and digital sovereignty that have long been of interest to Vitalik.The work is released under the GPL v3 license. Vitalik stated that adaptations into films, animations, and other derivative works are allowed, but requires the related production pipeline (including AI prompts, scripts, and other non-generic materials) to be open-sourced. He declared that all text was written by him, with AI assistance only used for spelling and grammar checks, consistency verification, and webpage formatting, and noted "Snowmoon was written with love."

The global fiber optic network project DoubleZero announced the launch of a dedicated market data source for the decentralized derivatives exchange Hyperliquid, allowing professional trading institutions to access its complete order book via fiber optics instead of public APIs. This data source covers the native perpetual contract market of Hyperliquid, as well as the markets operated by trade[XYZ] that run on its infrastructure, which offers perpetual contracts linked to assets such as crude oil, gold, and silver. This data source was co-developed by DoubleZero, validator node operators, and ecosystem partners Hyperion DeFi, MAVAN, and Kinetiq.DoubleZero stated that this service provides continuous and orderly market data streams for market makers, quantitative trading institutions, and proprietary trading firms, meeting their demand for faster and more stable order book updates. Previously, institutions wishing to obtain a complete view of Hyperliquid's order book had to integrate the data returned by public APIs themselves or run their own Hyperliquid nodes. DoubleZero noted that adjustments to Hyperliquid's public API have reduced the frequency and depth of available updates.Hyperliquid is the third trading venue to access the DoubleZero Edge market data service, following Solana and the prediction market Kalshi.

Two ChatGPT users from California filed a proposed class action lawsuit against OpenAI in the United States District Court for the Northern District of California this month, accusing the company of failing to adequately inform users that their real conversations were being handed over to external contractors for processing. The lawsuit was served to OpenAI on September 2, focusing on its internal initiative Project Lily. According to the complaint, "AI data reviewers" and "chatbot evaluators" recruited through a third-party staffing company read real ChatGPT prompts and complete conversations, summarize user intent, and score and comment on responses from four models on a scale of 1 to 7.This process is known in the industry as reinforcement learning from human feedback (RLHF), which is a fundamental method for enhancing chatbot capabilities. The complaint states that users were never explicitly informed that a person, rather than a machine, might be reading their conversations. OpenAI filters conversations through an automated system before human review, but the complaint alleges that the filters cannot intercept all content, and personal details sometimes still reach contractors. 404 Media first reported on the project on September 14 and found that the reviewers' dashboard included "user memory summaries," which could expose users' approximate locations, occupations, or private life information, even though usernames had been removed.OpenAI stated that such reviews aim to reduce two behaviors: chatbots behaving too much like humans and overly catering to users, referred to by researchers as "flattery." The complaint raises eight legal claims, including violations of California's Unfair Competition Law, Consumer Privacy Act, and common law claims for intrusion into private affairs, with the plaintiffs seeking damages, restitution of unjust enrichment, and punitive damages.

Social Capital founder Chamath Palihapitiya released an in-depth research report stating that open-source weight models are about four months away from matching the best closed-source frontier models in public evaluations, with increasing fluctuations in the gap. If open-source models allow companies more control over data, infrastructure, and customization while approaching frontier performance, the value corresponding to companies still paying for frontier laboratories becomes a business issue. Openness exists on a spectrum, from fully open-source models that can be downloaded and freely modified to open-source weight models with various restrictions, while closed-source models keep weights proprietary.Palantir CEO Alex Karp warned that companies might hand over differentiated proprietary knowledge and processes to frontier model providers. Microsoft CEO Satya Nadella stated that companies are effectively paying for intelligence twice: once in money and again in the more valuable proprietary knowledge that must be disclosed to make the intelligence useful. Despite concerns, companies are still willing to pay for frontier performance, even if the best open-source weight models are only months behind, with frontier laboratory revenues continuing to accelerate.Leading companies use both types of models, leveraging open-source models for control and customization while utilizing closed-source frontier models for maximum capability, with some cases reporting up to 12 times engineering efficiency and over 20 times cost savings. Some vendors adopt a dual-track approach, with Google offering both Gemini and Gemma, and Meta providing both Muse Spark and Llama. The 99-page report also discusses the costs of maintaining a lead for frontier laboratories, five factors of model competition, model operating locations, and investments in open-source weights by NVIDIA and Samsung.

A senior Iranian official stated that Qatar and Pakistan, as mediators, have informed Iran that the United States is ready to negotiate and reach an agreement, and that it is serious about advancing this process.The official pointed out that for Iran, as long as the United States returns to the Islamabad memorandum of understanding and fulfills its terms, Iran is also prepared to resume implementation of the memorandum within a few days. However, Iran demands practical and credible guarantees to ensure it will not be attacked again. Once the United States implements the substantive content of the memorandum, including stopping hostile actions, lifting the blockade, and allowing Iran to export oil, Iran will immediately initiate negotiations on nuclear issues and the lifting of sanctions.Iran has reached a consensus with Oman regarding the issue of the Strait of Hormuz; however, as long as the blockade continues and no agreement has been reached with the United States, "Iran will not reopen the strait."

According to Axios, multiple sources have revealed that even though artificial intelligence safety issues have become a hot topic of the era, Anthropic is still likely to go public in 2026, and the IPO schedule has not changed.

NVIDIA (NVDA.O) expands the open-source CUDA-Q platform for fault-tolerant quantum computing.

According to the Iranian Tasnim News Agency, Iran and Oman have reached an agreement after intensive consultations, and both sides have agreed on new route arrangements for entering and exiting the Strait of Hormuz.Informed sources say that this agreement does not mean the Strait of Hormuz is open; the southern route that the United States requested to be opened will be closed. Iran has proposed seven conditions for reopening to the United States, which may only be opened after receiving approval from the highest levels.

According to official news from Bitget, Bitget has completed a comprehensive upgrade of its U.S. stock product market data source, officially connecting to the Nasdaq official data source, becoming the first cryptocurrency trading platform in the industry to adopt this data standard. Compared to the previous data model, Nasdaq official data can provide users with more accurate and timely market data, and maintain better price stability and continuity during periods of significant market fluctuations.To further ensure data quality, after connecting to the official data, Bitget has established a multi-layered assurance mechanism based on the official metadata, including data processing, real-time monitoring, quality comparison, simulation validation, and anomaly data filtering. Nasdaq, as a core data service provider long adopted by global financial markets, has had its data standards validated over many years in the traditional financial sector.This upgrade also marks a further deepening of Bitget's global asset trading infrastructure development. As the trading categories gradually expand from digital assets to global assets such as U.S. stocks, the platform will continue to enhance its underlying data, liquidity, and risk management capabilities, and introduce more mature institutional-level market standards to provide individual investors and institutional clients with a more stable and efficient cross-market trading experience, continuously enhancing the comprehensive competitiveness of its Universal Exchange (UEX).

According to The Block, ZKsync developer Matter Labs has open-sourced the permission engine of its distributed ledger platform Prividium for financial institutions, allowing organizations to run, inspect, and modify core code in their own environments. The German central bank has become the first institution to test and deploy Prividium.Matter Labs CEO Alex Gluchowski stated that feedback from regulators regarding commercial core means reliance on a single vendor, and that open-sourcing the core "structurally eliminates adoption barriers."Other components of Prividium, such as the ZKsync OS core and Atlas sequencer, have previously been open-sourced. The permission engine is planned to operate as an independent component, allowing institutions to deploy it without a commercial agreement, but management tools and integration with existing systems remain commercial products. Gluchowski said, "The central bank running public code is the clearest signal in this direction."

Supply chain sources revealed that due to the overall cost surge, Intel's PC CPUs have been gradually increasing in price over the past year, with a tentative additional increase of 10% expected in early October.In the context of a slight decline projected for the PC terminal market in 2027, Intel still chooses to raise CPU prices. Supply chain analysis indicates that if Intel sets low profit margins as an important threshold for product retention, some products will inevitably be forced to exit the market. Once this demand is released, it may provide opportunities for Arm-based operators like MediaTek and Qualcomm to enter, especially in the IPC, edge computing, and IoT markets, where Arm SoCs have advantages in high integration and low power consumption.

Starknet lending protocol Vesu announced that an oracle incident occurred on September 4. A malfunctioning upstream Pragma price source led to the abnormal liquidation of 47 positions and $3 million in collateral across multiple Vesu liquidity pools between 04:08 and 04:10 UTC. The price source self-corrected within two minutes and has not been affected since.Pragma has collaborated with relevant parties to deploy a fix for the root cause, and Vesu expects liquidity pool curators to lift the suspension of the pools in the coming hours. Vesu stated that its contracts are operating as designed, the protocol has no vulnerabilities, and no fixes need to be deployed; the issue was due to incorrect input prices. Vesu is working with Pragma, StarkWare, Starknet Foundation, and liquidity pool curators to recover the funds taken through these liquidations. Affected users should keep their Earn positions open to avoid losing refund eligibility; if liquidated during the incident, please submit a support ticket on Discord. A complete technical report will be released soon.

Base Labs has announced its establishment as an independent research organization focused on advancing open-source AI. It believes in a healthy and open frontier model ecosystem and will conduct scientific research in an open manner.Base Labs will carry out blue-sky research on continuous learning, reinforcement learning science, and model lifecycle learning mechanisms, and will publicly share all experiments and recipes; it will build the BaseHub Data Foundry to provide an open reinforcement learning environment, training data, and real-world benchmarks; it will enhance the capabilities, safety, and alignment of open-source models through continuous post-training and deploy cutting-edge safety stacks; at the same time, it will conduct model performance research to reduce costs and improve performance.The organization emphasizes that this is mission-driven research work rather than a commercial product and is currently recruiting engineers, researchers, and research scholars.

NVIDIA announced that the RTX Spark devices equipped with the N1X chip will begin shipping in October this year. The N1X supports up to 128GB of unified memory, and the Blackwell GPU provides up to 10 petaflops of floating-point performance per second. The full version is equipped with a 6144 CUDA core Blackwell GPU and a 20-core Grace CPU, supporting 24GB to 128GB of unified memory; another version has 5120 GPU cores and 18 CPU cores, supporting only 24GB to 32GB of unified memory.NVIDIA launched the open-source tool NVIDIA PAIR, which can connect multiple RTX devices, DGX Spark, and even Apple devices in a home network to schedule idle computing power for collaborative processing of AI agent tasks. Compatible devices include computers with NVIDIA graphics cards (RTX 20 series and later, RTX Pro GPU, DGX Spark) as well as devices with Apple M4 or newer chips. PAIR prioritizes the use of idle computing power and automatically adjusts as devices join or leave the network.In a media briefing example, a household had approximately 165 TFLOPS of underutilized computing power. The Qianwen 3.6 35B A3B model completed agent tasks in an average of 18 minutes on a single Spark notebook, while a three-device PAIR cluster averaged 8 minutes and 48 seconds. The AI agent applications Perplexity Portable Computer, Hermes Agent, and OpenClaw will receive a more simplified local deployment method.