Solana Co-Founder: Solana v2.2 Delivers 20% Increase in Compute Capacity, Equivalent to Adding 5 Ethereum Pectra Upgrades' Worth of Capacity to the Network

Beating AI Insight News Brief: Claude Code v2.1.295 now officially supports OSC 7501, a terminal state protocol recently proposed by Ghostty creator Mitchell Hashimoto. The protocol lets Claude Code actively notify the terminal of its status—working, waiting for the user, or completed—removing the need for external tools to infer status from screen content. The period between the protocol’s announcement and Claude Code’s release was roughly 48 hours; notably, Claude Code was exactly the example Mitchell initially used to demonstrate the protocol. Previously, Herdr relied on 16 identification rules to judge Claude Code’s status, revising them around 10 times over the past three months, with even a single character change in Claude Code’s loading animation requiring full rule re-adaptation. Now, status is reported directly by Claude Code itself. As long as the terminal supports OSC 7501, users can immediately confirm if Claude Code is working, waiting for user input, or finished. Mitchell added that the Anthropic team has verified this implementation using Rex, which currently has preliminary support for the protocol.

CryptoQuant analyst Darkfost noted in a post that despite the crypto market entering a short-term correction recently, Ethereum’s price has fallen roughly 11% and its market capitalization has shrunk by over $38 billion, investors are still steadily withdrawing ETH from trading platforms. Data shows that Binance’s ETH reserves dropped from 3.57 million to 3.47 million over the past three days, hitting a near six-month low. Since August, Binance’s ETH reserves have declined by about 11.5% from 3.92 million. Meanwhile, the number of Binance ETH withdrawal transactions surpassed 320,000 on October 6, setting a new all-time high.

US President Trump said he will not launch airstrikes on Iran ahead of next month’s midterm elections, easing market concerns over escalating tensions in the Middle East. In response, Bitcoin rebounded from its Thursday low of around $80,400 to near $82,500, putting short sellers under liquidation pressure. In terms of market prices, Bitcoin is currently trading at $82,486; Ethereum fell roughly 2% in 24 hours to near $2,500; SOL dropped nearly 4% to around $110, while XRP and HYPE remained relatively stable, with the CoinDesk 20 Index down approximately 2%. Alex Kuptsikevich, chief market analyst at FxPro, noted that Bitcoin dipped to $80,400 on Thursday, hitting a nearly three-week low and its 50-day moving average before securing buying support, indicating sustained buying demand during the pullback. In broader macro markets, S&P 500 futures rose less than 0.5%, while Nasdaq 100 futures gained nearly 1%; Brent crude oil fell below $103 per barrel, easing inflation concerns stoked by earlier oil price increases. Whether Bitcoin can stage a further rebound will depend on its upside resistance levels and developments in the Middle East.

According to CNBC, OpenAI and Anthropic are recruiting former Trump administration officials for senior roles to strengthen their ties with the U.S. government. As advanced AI technology is increasingly viewed as a national security issue, both companies are ramping up their political affairs operations to address policy impacts including regulation, federal government procurement, supply chain risk designations, and export controls. So far this year, Anthropic has hired at least two former Trump administration officials: former White House Deputy Chief of Staff Chris Liddell joined its board in February, and Sihao Huang, a former senior policy advisor at the White House Office of Science and Technology Policy (OSTP), joined the company in September to lead its advanced computing strategy. OpenAI has also recently hired several former government officials, including Thomas Lind (who oversaw AI policy at the White House Office of the National Cyber Director), Dean Ball (who helped draft the Trump administration’s “American AI Initiative”), and Peter Bowman-Davis (former chief AI officer at the U.S. Department of Health and Human Services). Anthropic previously had a public conflict with the U.S. Department of Defense, and in September, a U.S. appeals court upheld its supply chain risk designation. Since then, Anthropic CEO Dario Amodei has recently met with Trump and other AI industry executives to try to ease tensions with the administration. Analysts note that as AI companies face increasingly complex regulatory and national security requirements, government relations have become a key part of industry competition. Both companies are currently hiring multiple political affairs-related roles to gain greater influence in U.S. AI policy-making.

Insight Beating AI News: AI model leaderboard Arena has completed a $200 million Series B funding round, valuing the company at $3.1 billion. The round was co-led by Lightspeed Venture Partners and Khosla Ventures, with participation from Salesforce Ventures, a16z, and other investors. Back in January this year, Arena had just closed a $150 million Series A round at a $1.7 billion valuation. Originating from a research project at the University of California, Berkeley, Arena initially let users anonymously compare responses from different AI models and generate leaderboards via voting. Now, the company is rolling out paid evaluation and analysis services for model providers and enterprises. Arena reported its annual recurring revenue (ARR) has grown from $30 million in January to $100 million in June. Currently, the platform draws tens of millions of monthly visitors. Official disclosures show the platform has accumulated 350 million sessions and 62 million votes to date, covering tasks including text, programming, image, and video. Alongside the funding, Arena also launched a new AI Agent security leaderboard, the Alignment Index. This leaderboard analyzed over 90,000 real sessions across 27 models, checking for unauthorized operations, misinterpretation of user requests, or false task completion reports by AIs. OpenAI models took all top five spots, with GPT-6.1 Sol leading at 87.9 points. The study also found that 48% of code debugging sessions involved false completion claims.

Polychain Capital founder and CEO Olaf Carlson-Wee announced at the Network State Conference that Polychain is leading Network School’s new financing round, which values the company at $2 billion. Participating investors include Andreessen Horowitz (a16z), Coinbase, and Balaji himself, among other institutions and individuals; the specific amount raised was not disclosed. Carlson-Wee noted that Polychain’s investment strategy does not focus solely on incremental improvements, but rather targets projects that leverage new business models or technologies to solve problems in innovative ways. He expressed anticipation for participating in Network School’s funding round, adding that he has known and collaborated with Balaji for over a decade.