OpenAI unleashes its ultimate weapon.

In the early hours of Friday, Beijing time, OpenAI officially released GPT-6 Astra, calling it "the world's most intelligent model with the highest level of alignment." In response, Rich Privorotsky, head of Goldman Sachs' Delta-One trading desk, published a report over the weekend highlighting Astra's importance to the current development of the AI industry. On Friday, in the U.S. stock market, SoftBank saw its stock price surge nearly 8% due to optimistic sentiment surrounding its stake in OpenAI, while Oracle also rose more than 3%. However, industry insiders generally believe that the surge in stock prices is actually the least noteworthy part of the entire story compared to the deeper underlying logic. Interestingly, the performance of the OpenAI camp this week has significantly overshadowed that of the Anthropic camp... Privorotsky noted that the true excitement in the market lies in the emergence of a leading large modelprecisely what the bullish assertions about AI have been anticipating. With Astra's impressive performance in AGI (Artificial General Intelligence) benchmark tests, OpenAI finally has the confidence to declare that it has fully surpassed Anthropic on several core metrics. As OpenAI President Greg Brockman stated on Friday, Astra represents "a leap for a generation" and indicated that it might ultimately be viewed as a landmark moment for the arrival of AGI. When asked whether Astra meets this definition, Brockman replied, "I think this might be that model." Although there is currently no widely accepted standard for AGI testing, and Brockman has stated that the final judgment will be left to users, he nonetheless concluded his conversation with reporters by saying, "Welcome to the AGI era." Is the perspective on AI finally shifting back from cost reduction to technological breakthroughs? Privorotsky pointed out that while the lower usage cost threshold in the AI field is undoubtedly a good thing, it is the substantial breakthroughs in underlying capabilities that truly steal the spotlight. As Privorotsky explained in his latest report: When models become genuinely more intelligent, businesses will not simply use the same amount of AI at a lower price. They will continuously uncover entirely new application scenarios, leading to a reconfiguration of the entire demand landscape. This will compel other competing labs to keep up, preventing the spending cycle from stagnating and reigniting belief in goals still worth pursuing. Interestingly, earlier this week, Privorotsky had also focused on the decline in current token spending in the AI sector, noting that the growth in token demand has not kept pace with the declining cost of tokens, which led to some confusion in the industry about his stance being pessimistic. He had stated at the time, "If the decline of tokens outpaces the growth of consumption, then increased demand + falling prices does not necessarily indicate a bullish situation. Unless AI capabilities (like the latest Astra) continue to achieve breakthrough advancements... otherwise, a phase of oversupply in computing resources may occur." In his latest report, Privorotsky further clarified that he is by no means bearish on AIalthough he admitted that the hardware sector's profitability valuations are indeed declining, the underlying logic behind this is worth pondering. But as he stated, the emergence of Astra precisely points in the entirely opposite direction for this core topic: ...breakthroughs in intelligence are key to reigniting the market's willingnessand necessityto invest continuously. In the conclusion of his report, Privorotsky also noted that the schedule for September seems to favor the AI sector rather than being a negative influence. Oracle will announce its earnings next week, the analyst meeting schedule is busy, and there are many opportunities for bullish signals to propagate throughout the entire industry... This article is reproduced from Financial Alliance," author: Xiaoxiang, GMTEight editor: Xu Wenqiang.