14x Speed Rush! OpenAI Launches GPT-5.6 Sol UltraFast Super Fast Mode, Up to 750 Tokens Per Second

OpenAI has officially launched the new GPT-5.6Sol UltraFast super-fast mode preview for enterprise users. With this mode, the model's performance has seen a revolutionary improvement, reaching up to 14 times the speed of the standard mode, and can generate up to 750 tokens per second.

According to the information, the core technology of this ultra-fast mode is supported by Cerebras. However, in order to ensure the reasonable use of resources, this mode is currently available through an application review system. Enterprise users who need it must submit specific usage scenarios to OpenAI, and whether the application is approved will be comprehensively evaluated by the official based on actual circumstances.

In terms of application scenarios, OpenAI stated that the ultra-fast mode is not aimed at all regular workloads, but is specifically designed for business scenarios that require high real-time performance. For example: real-time voice interaction, intelligent customer support, business applications, development of intelligent agents (Agents), financial in-depth research, and cutting-edge security research.

As the preview phase progresses, the computing power provided by Cerebras will continue to be optimized. In the future, OpenAI plans to gradually expand the coverage based on feedback from enterprise workloads, allowing more enterprises to smoothly access this ultra-fast AI capability through the API interface.

Sam Altman, CEO of OpenAI: It's Too Long to Study at University for Four Years, Two Years Are Enough

Sam Altman said in an interview that it was just right to study for two years and drop out of university. He believes that two years at Stanford have already provided him with the core knowledge needed for entrepreneurship, and spending another two years would bring little additional value and low cost-effectiveness. In today's rapidly evolving technological environment, this growth path may also suit many other young people.

OpenAI Dissolves Security Team, Security Functions Merged into Existing Business Departments

OpenAI disbanded its preparedness team late last month, FT reported Aug 17. The team assessed frontier model risks such as bio/chemical and cyberattacks and devised mitigations; duties were split by risk area into business teams. This follows disbanding of AGI readiness and superalignment teams, showing a shift from safety-oriented public-interest governance.....

NVIDIA is reportedly planning to invest 3 billion USD in SB Energy, a subsidiary of SoftBank, to further support the OpenAI data center ecosystem

NVIDIA is discussing with SB Energy, a clean energy company supported by SoftBank, to invest up to 3 billion USD as a core part of its participation in the OpenAI data center transaction. The move aims to accelerate the construction of AI infrastructure and deeply support the AI industry chain with substantial financial power.

OpenAI Annual Revenue Exceeds $40 Billion, Surges Over 20% MoM in July

OpenAI is accelerating its plan to go public in the US this year, with its valuation expected to exceed $1 trillion. Internal documents show that its annualized revenue has surpassed $40 billion, doubling since the end of last year; revenue in July grew by more than 20% compared to the previous month, showing strong financial performance and attracting high attention from the capital market.

ChatGPT for Mac Launches Computer History: Say Goodbye to Screenshot Records and Build Your Work Memory with System Events

OpenAI launched Computer History on Aug 13, optionally replacing Chronicle preview, adding activity memory and timeline to ChatGPT for Mac. It uses macOS accessibility APIs, not screenshots, to record clicks, typing, shortcuts, and app switches, turning them into structured memory for ChatGPT and Codex to generate work summaries.....