// NODE-02 · NEURAL FEED · DAILY TRANSMISSION //
July 14, 2026 · WINDYVIEWS.COM
OpenAI moved its GPT-5.6 lineup — Sol, Terra, and Luna — from a two-week government-gated preview to broad availability on July 9, making it ChatGPT's default and shipping alongside a new ChatGPT Work desktop app. The company says the flagship Sol tier hits state-of-the-art results in coding, cybersecurity, and science using fewer tokens, while the balanced Terra tier matches the older GPT-5.5 at roughly half the cost.
xAI released Grok 4.5, positioning it around stronger coding, agentic task handling, and general knowledge work. It landed in the same busy week as OpenAI's GPT-5.6 broad launch, Meta's first paid model, and a sub-$2 coding model from Cognition — a sign of how compressed the frontier release cadence has become.
Anthropic returned Claude Fable 5 to all users worldwide on July 1 after the US Commerce Department lifted export controls it had imposed in June, and extended Fable 5 to every paid plan. It also kept Claude Code's weekly rate limits 50% higher through July 19 and added a built-in sandboxed browser to the Claude Code desktop app, letting the assistant pull up and interact with live websites.
South Korean chipmaker SK Hynix began Nasdaq trading on July 10 in the largest American Depositary Receipt offering in history, raising roughly $26.5 billion. As the dominant supplier of the high-bandwidth memory that Nvidia's AI accelerators depend on, its stock has climbed about 650% over the past year.
The New York Times and New York Daily News filed a sanctions motion against OpenAI on July 9, alleging the company concealed tools capable of identifying copyrighted journalism in ChatGPT outputs. Court-ordered depositions reportedly revealed OpenAI held 78 million de-identified chat logs, which the plaintiffs say contradicts its earlier claims.
Israeli healthcare researchers demonstrated an AI system that detects life-threatening brain hemorrhages on CT scans in seconds — before physicians can visually identify them — and alerts clinical teams in real time. The work points to a growing role for AI as a first-pass triage layer in emergency imaging.
A London workshop backed by Logos Research is using AI autoformalization tools to translate Fermat's Last Theorem into a machine-verifiable proof format. The project is an early test of whether AI can meaningfully accelerate the formalization of advanced mathematics rather than just assist at the margins.
New York City Schools Chancellor Kamar Samuels told principals on July 8 to pause all educational software purchases until the city finalizes its AI policy. An earlier draft drew nearly 6,500 public comments and prompted 29 City Council members to call for a two-year moratorium — a reminder that AI adoption in public institutions is running into real friction.
The frontier-model race has gone from months to weeks: GPT-5.6, Grok 4.5, Claude Fable 5/Sonnet 5, Meta's first paid model, and a sub-$2 coder all shipped inside one week, and the pitch has shifted from raw size to cost, efficiency, and reliable task completion. Money is chasing the picks-and-shovels layer just as hard, with SK Hynix's record $26.5B memory-chip listing. But the friction is just as real — copyright litigation against OpenAI is intensifying and institutions like NYC's schools are hitting pause — even as AI keeps notching genuine wins in medicine and mathematics.