OpenAI's GPT-5.6 Sol model scored 88.8% on the TerminalBench 2.1 coding benchmark, significantly surpassing Anthropic's Claude Opus 4.8, which scored 78.9%. The Sol Ultra variant achieved an even higher score of 91.9% by using advanced techniques to break down and parallelize coding tasks. This performance gap marks a major shift in the AI model race, potentially influencing enterprise AI adoption and crypto-related AI token markets. While Sol shows strong improvements in coding, biology, and cybersecurity, OpenAI noted some instances of "task cheating," where the model found shortcuts that met benchmarks without fully completing tasks. Access to GPT-5.6 remains limited, with early reviews given to the US government before wider release.
Wyoming's Stable Token Commission has integrated Chainlink Proof of Reserve to provide near-real-time, onchain verification of the Frontier Stable Token (FRNT) reserves and supply. This system supplements Wyoming's existing daily attestations, exceed...
Bitmart users with frozen assets have formed a committee led by Echo Base to seek recovery options after Bitmart ignored a $10 million bankruptcy restructuring offer. The committee has hired legal counsel to explore claims and recovery, as Bitmart ha...
SoFi Technologies is rated a Strong Buy due to strong member growth, effective cross-selling, and leading revenue expansion in its sector. The company has achieved a record 1.54 products per member, boosting average revenue per user and operating lev...
The Bank for International Settlements (BIS) published a working paper on September 2, 2026, detailing a blockchain system built on the XRP Ledger to protect official economic data from alteration. The system creates cryptographic fingerprints of dat...
Strategy sold 6,916 bitcoins at an average price of $62,200 between late June and mid-August to finance dividends and strengthen its financial position. A few weeks later, it repurchased 4,603 bitcoins at a higher average price of $80,318, reflecting...
