Crypto News & Market Updates

BlockBeats reported on June 5 that according to Saudi Arabia's Al Arab TV, Iran has notified Pakistan that Iran has accepted the transfer of part of its uranium stockpile to a third country it has agreed to. (Jin Shi)

According to Bijie, Circle's CEO provided detailed answers to the 13 most pressing questions from investors in the latest AMA. Circle regards ARC and Agent Stack as essential infrastructure for internet finance platforms, focusing on AI proxy payments and machine-to-machine payment settlement. USDC's slowdown in growth is partly affected by macro capital rotation, but institutional consolidation, internationalization, and real-world use cases are still progressing. Circle emphasizes that network effects, interoperability, liquidity, and regulatory compliance form a long-term moat, with ARC, CCTP, StableFX, and others being key components of its on-chain financial infrastructure.

According to Bijie.com, the U.S. spot Bitcoin ETF recorded a small net inflow of $3.05 million on Wednesday, ending a 13-day outflow that had lasted since mid-May, during which over $4.4 billion was outflowed, reducing the total assets of Bitcoin ETFs from $104.29 billion to $80.4 billion. Blackstone's IBIT absorbed $47.66 million, while Fidelity's FBTC, Bitwise's BITB, and ARK's ARKB continued to flow out. According to Checkonchain data, the total assets under management for U.S. spot Bitcoin ETFs are 1.277 million BTC, slightly higher than the low of 1.274 million BTC on February 23. After a 17-day outflow spree, spot Ethereum ETFs recorded a net inflow of $19.3 million on Wednesday, all from Blackstone's ETHA, with no net inflows from other Ethereum ETFs. The total assets of Ethereum ETFs are $9.78 billion, accounting for 4.57% of Ethereum's circulating market capitalization.

According to ChainCatcher, decentralized AI infrastructure network DGrid AI has announced an official partnership with MiniMax. Through the DGrid decentralized AI Gateway, both parties will provide global developers and agents with low-cost, high-availability access to the MiniMax model, jointly promoting its adoption across the broader ecosystem. DGrid is a decentralized AI infrastructure that aggregates mainstream global AI models, currently integrating 200+ models including Claude, GPT, and Gemini, enabling one-stop calls through a unified API. The platform targets developers and AI Agent builders, offering high-availability, low-latency model access services, aiming to lower the entry barrier and usage costs for AI applications. MiniMax is a global leader in full-modal general artificial intelligence, and one of the few AI companies with self-developed capabilities in text, speech, video, music, and other full-modal models. This official collaboration with MiniMax marks another important expansion of DGrid on the model supply side, marking a new step forward for the platform in connecting mainstream model vendors and building a decentralized AI ecosystem. With the integration of more high-quality models, DGrid is becoming a unified gateway for AI developers to access diverse model capabilities, providing more flexible and open infrastructure support for next-generation intelligent applications. As the first batch of collaborations, the MiniMax flagship model M3 has launched on DGrid, and DGrid Premium users can enjoy a direct 45% discount.

According to Bijie.com, Google DeepMind has released the open-source real-time music generation model Magenta RealTime 2 (MRT2). The new model is optimized specifically for Apple (M series) MacBooks, supporting local ultra-low latency operation. Users can play and control the model in real time as an instrument using MIDI signals, text prompts, or audio clips with a delay of less than 200 milliseconds. The model supports multiple interactive methods to guide music generation, including MIDI guided mode, text-to-synthesizer mode, and audio cloning mode. To meet the needs of different hardware configurations, the development team released two versions with different specifications: the MRT2_small version with 230 million parameters and the MRT2_base version with 2.4 billion parameters. The inference library Magenta-RT and the C++-based inference engine Magentart::core are both open-sourced on GitHub. The official team also provides standalone applications and digital audio workstation (DAW) plugins that run directly on macOS.

According to BlockBeats news on June 5, according to Onchain Lens monitoring, a whale holding 38,554 ETH (about $64.28 million) reactivated after three years of dormancy. He deposited 20,000 ETH (about $33.28 million) into Aave V3 as collateral to lend 30 million USDT, then bought 17,826 ETH at an average price of $1,683 (through revolving borrowing and leverage), currently holding a total of 56,380 ETH (approximately $94.04 million).

Odaily Planet Daily reports that driven by a sharp drop in the Korean stock market, emerging market stocks and currencies have weakened for the third consecutive trading day. The MSCI Emerging Markets Index fell 1.7% to 1,728.66 points that day, while the Korea Composite Index (Kospi) once fell as much as 7% during trading. This sell-off was mainly concentrated in the chip sector. Samsung Electronics and SK Hynix led the decline, together accounting for more than half of Kospi's market value. Previously, Broadcom's announced AI chip sales outlook failed to meet market expectations, triggering a shift in investor sentiment. Charu Chaana, Chief Investment Strategist at Saxo Markets, said that if U.S. nonfarm payroll data is stronger than expected, it could become another reason for investors to cut back on crowded AI deals. The market is currently closely watching this data to assess the Fed's future interest rate path.

BlockBeats reported on June 5 that the latest research from the Boston Fed points out that as energy efficiency improves and domestic crude oil production increases, the U.S. economy's sensitivity to rising oil prices has significantly decreased. Unlike the oil crisis of the 1970s, today's oil price increases no longer have a large-scale impact on the job market. The new jobs created by oil and gas industry expansion can partially offset the pressure on other industries, making it much less likely for high oil prices to trigger stagflation caused by 'high inflation + high unemployment.' However, the report also warns that the oil price shock weakens its buffer mechanism for employment, meaning inflationary pressures from rising energy prices may become more prolonged. The Fed no longer needs to worry excessively about energy price increases leading to a recession, but should instead focus its policy on curbing inflation. Currently, the market generally expects the Fed to keep rates unchanged at the June meeting, but some officials have begun discussing the possibility of a rate hike within the year. Meanwhile, Morgan Stanley believes that this round of oil price increases is more of a short-term supply disturbance and not enough to be the core driver for rate hikes. The firm expects U.S. interest rates to remain unchanged throughout the year and is expected to start a rate-cutting cycle in 2027. However, as geopolitical conflicts have driven up energy prices, the market has clearly shifted its policy path. Fed officials have recently frequently sent hawkish signals, emphasizing that if inflation remains above targets, further tightening is still possible. (Jin Shi)

According to Bijie.com, LM Studio has officially launched its official mobile app Locally (supports iPhone and iPad), deeply integrating it into its own ecosystem. The app supports "On-device Local Run" and "LM LINK" functions, allowing users to remotely connect to desktop computing power and break the physical boundaries of device computing power. Locally optimized for Apple Sil's early releases, based on the MLX machine learning framework, supporting mainstream open-source models such as Llama 3.2, Google Gemma 4 series, Qwen 3, and DeepSeek. Users can run Gemma 4 (E2B) models directly on their phones and import custom GGUF format models. Locally insists on 100% offline and privacy protection, with all processing completed locally on the device. To enable LM LINK, users need to log in to the same LM Studio account on both mobile and computer for authorization, establishing a Tailscale-based encrypted network. Users can securely connect to desktop computers at home or office to remotely access large-parameter models.