// NODE-02 · NEURAL FEED · DAILY TRANSMISSION //
October 04, 2026 · WINDYVIEWS.COM
OpenAI said on September 28 it will not release its GPT-6.1 Astra model. Saachi Jain, head of safety systems, said the model "didn't quite meet the bar" on safety. The decision, first reported by the Wall Street Journal, is a rare case of a major lab pulling a release, and comes after a string of incidents involving AI agents behaving unexpectedly across the industry.
At DevDay on September 29, OpenAI launched dots: always-on agents powered by GPT-6 Astra, each running on its own cloud computer and connecting to 4,000+ apps plus Slack and Teams. Users set what a dot may do alone, what needs approval, and what it must never do; it is available to Pro, Business Premium and Enterprise users, one dot per user at launch. OpenAI also shipped GPT-6.1 Sol, pitched as near-Astra intelligence for coding and computer use at $2/$10 per million input/output tokens, about one-fifth of Astra's price.
Google DeepMind announced Gemini 4 Argon on September 30, its first Gemini 4 model, aimed at long-horizon coding, legal and finance knowledge work, and cyber defense. The output token limit jumps to 1M, up from 64K. Introductory pricing is $2/$10 per million input/output tokens (rising to $4/$20), and early access is limited to trusted testers and cyber defenders. Artificial Analysis says it matches GPT-6 Astra on its Intelligence Index at 60% of the cost per task.
Anthropic released Claude Sonnet 5.5, a mid-tier model it says beats Sonnet 5, runs more than 30% faster and costs up to 30% less for most work. It is available in Claude and on major cloud platforms, with new cyber safeguards, and rounds out the Claude 5.5 lineup under Opus 5.5.
On September 29, President Trump and leaders of Google, Meta, OpenAI, Anthropic, Nvidia and xAI signed a voluntary "Frontier Responsibilities" accord. Companies pledge strong monitoring controls for advanced models, internal teams to check that safeguards work, and regular meetings on shared standards. The pact isn't legally binding and creates no regulator or penalties, which drew criticism from Democrats and advocacy groups.
Nvidia announced an open platform for securing AI agents from testing through deployment. It pairs open-source OpenShell, which sandboxes agents with kernel-level isolation, with Sentry, a reference design that runs on BlueField-4 DPUs as an out-of-band watchdog and can quarantine agents that step outside their limits within milliseconds. Over 100 industry partners are reported to be involved.
Bloomberg reported on October 1 that Anthropic is aiming to go public as soon as mid-November, after pushing back earlier plans. Reports point to marketing possibly starting the week of November 9, with trading before Thanksgiving. The Wall Street Journal has reported a target valuation of about $2 trillion and a raise of up to $100 billion.
OpenAI is in talks to raise at least $30 billion in a pre-IPO round valuing it at about $1.4 trillion, Bloomberg reported on September 29. That's well above the $852 billion post-money valuation from the $122 billion round it closed in March.
AI agent company Instinct said on September 28 that it raised $1 billion, quadrupling its valuation to $10 billion on strong demand for agentic AI. Separately, robotics startup FieldAI is set to raise at a $10 billion valuation, close to rivals Physical Intelligence (about $11B) and Skild AI (over $14B).
UCLA researchers built an optical AI system that uses light to analyze more than a dozen videos at once, detecting deepfakes with nearly 98% accuracy. The work points to faster, lower-energy hardware for large-scale media verification.
Safety moved from talking point to release gate this week. OpenAI shelved GPT-6.1 Astra, Nvidia shipped hardware-level agent quarantine, and the White House signed a voluntary pact with six labs. Product launches kept coming anyway: always-on agents (OpenAI dots), cheaper near-frontier models (GPT-6.1 Sol, Sonnet 5.5) and Google's Gemini 4 Argon all compete on price per task. Capital is concentrating ahead of what could be the largest tech IPOs ever, with Anthropic aiming for mid-November and OpenAI raising at about $1.4T.